diff --git a/assets/css/markdown.css b/assets/css/markdown.css
index bbaba8507..b9511d418 100644
--- a/assets/css/markdown.css
+++ b/assets/css/markdown.css
@@ -5,17 +5,29 @@
 }
 
 .markdown h1 {
-  @apply text-gray-900 font-semibold text-3xl my-4;
+  @apply text-gray-900 font-semibold text-4xl my-4;
 }
 
 .markdown h2 {
-  @apply text-gray-900 font-semibold text-2xl my-4;
+  @apply text-gray-900 font-semibold text-3xl my-4;
 }
 
 .markdown h3 {
+  @apply text-gray-900 font-semibold text-2xl my-4;
+}
+
+.markdown h4 {
   @apply text-gray-900 font-semibold text-xl my-4;
 }
 
+.markdown h5 {
+  @apply text-gray-900 font-semibold text-lg my-4;
+}
+
+.markdown h6 {
+  @apply text-gray-900 font-semibold text-base my-4;
+}
+
 .markdown p {
   @apply my-4;
 }
@@ -97,3 +109,16 @@
 .markdown :last-child {
   @apply mb-0;
 }
+
+/* Overrides for user-entered markdown */
+
+.cell .markdown h1,
+.cell .markdown h2 {
+  font-size: 0;
+}
+
+.cell .markdown h1:after,
+.cell .markdown h2:after {
+  @apply text-red-400 text-base font-medium;
+  content: "warning: heading levels 1 and 2 are reserved for notebook and section names, please use heading 3 and above.";
+}
diff --git a/lib/live_book/live_markdown.ex b/lib/live_book/live_markdown.ex
new file mode 100644
index 000000000..6b703322f
--- /dev/null
+++ b/lib/live_book/live_markdown.ex
@@ -0,0 +1,50 @@
+defmodule LiveBook.LiveMarkdown do
+  @moduledoc false
+
+  # Notebook file format used by LiveBook.
+  #
+  # The format is based off of Markdown and preserves compatibility,
+  # in the sense that every LiveMarkdown file is a valid Markdown file.
+  # LiveMarkdown uses HTML comments for storing metadata, so a Markdown standard
+  # supporting that is assumed. Not every Markdown file is a valid LiveMarkdown file,
+  # but may be converted to such by applying tiny changes, which the import function does.
+  #
+  # Currently the format is straightforward and specifies the following:
+  #
+  # 1. The file should have a leading *Heading 1* holding the notebook name.
+  # 2. Every *Heading 2* starts a new section.
+  # 3. Every Elixir code block maps to an Elixir cell.
+  # 4. Adjacent regular Markdown text maps to a Markdown cell.
+  # 5. Comments of the form `<!--live_book:json_object-->` hold metadata
+  #    any apply to the element they directly precede (e.g. an Elixir cell).
+  #
+  # ## Example
+  #
+  # Here's an example LiveMarkdown file:
+  #
+  #     # My Notebook
+  #
+  #     ## Section 1
+  #
+  #     Make sure to install:
+  #
+  #     * Erlang
+  #     * Elixir
+  #     * PostgreSQL
+  #
+  #     <!--live_book:{"readonly":true}-->
+  #     ```elixir
+  #     Enum.to_list(1..10)
+  #     ```
+  #
+  #     This is it for this section.
+  #
+  #     ## Section 2
+  #
+  #     ```elixir
+  #     # More Elixir code
+  #     ```
+  #
+  # This file defines a notebook named *My Notebook* with two sections.
+  # The first section includes 3 cells and the second section includes 1 Elixir cell.
+end
diff --git a/lib/live_book/live_markdown/export.ex b/lib/live_book/live_markdown/export.ex
new file mode 100644
index 000000000..4748ca17a
--- /dev/null
+++ b/lib/live_book/live_markdown/export.ex
@@ -0,0 +1,81 @@
+defmodule LiveBook.LiveMarkdown.Export do
+  alias LiveBook.Notebook
+  alias LiveBook.LiveMarkdown.MarkdownHelpers
+
+  @doc """
+  Converts the given notebook into a Markdown document.
+  """
+  @spec notebook_to_markdown(Notebook.t()) :: String.t()
+  def notebook_to_markdown(notebook) do
+    iodata = render_notebook(notebook)
+    # Add trailing newline
+    IO.iodata_to_binary([iodata, "\n"])
+  end
+
+  defp render_notebook(notebook) do
+    name = "# #{notebook.name}"
+    sections = Enum.map(notebook.sections, &render_section/1)
+
+    [name | sections]
+    |> Enum.intersperse("\n\n")
+    |> prepend_metadata(notebook.metadata)
+  end
+
+  defp render_section(section) do
+    name = "## #{section.name}"
+    cells = Enum.map(section.cells, &render_cell/1)
+
+    [name | cells]
+    |> Enum.intersperse("\n\n")
+    |> prepend_metadata(section.metadata)
+  end
+
+  defp render_cell(%{type: :markdown} = cell) do
+    cell.source
+    |> format_markdown_source()
+    |> prepend_metadata(cell.metadata)
+  end
+
+  defp render_cell(%{type: :elixir} = cell) do
+    """
+    ```elixir
+    #{cell.source}
+    ```\
+    """
+    |> prepend_metadata(cell.metadata)
+  end
+
+  defp render_metadata(metadata) do
+    metadata_json = Jason.encode!(metadata)
+    "<!--live_book:#{metadata_json}-->"
+  end
+
+  defp prepend_metadata(iodata, metadata) when metadata == %{}, do: iodata
+
+  defp prepend_metadata(iodata, metadata) do
+    content = render_metadata(metadata)
+    [content, "\n", iodata]
+  end
+
+  defp format_markdown_source(markdown) do
+    markdown
+    |> EarmarkParser.as_ast()
+    |> elem(1)
+    |> rewrite_ast()
+    |> MarkdownHelpers.markdown_from_ast()
+  end
+
+  # Alters AST of the user-entered markdown.
+  defp rewrite_ast(ast) do
+    ast
+    |> remove_reserved_headings()
+  end
+
+  defp remove_reserved_headings(ast) do
+    Enum.filter(ast, fn
+      {"h1", _, _, _} -> false
+      {"h2", _, _, _} -> false
+      _ast_node -> true
+    end)
+  end
+end
diff --git a/lib/live_book/live_markdown/import.ex b/lib/live_book/live_markdown/import.ex
new file mode 100644
index 000000000..894e47668
--- /dev/null
+++ b/lib/live_book/live_markdown/import.ex
@@ -0,0 +1,187 @@
+defmodule LiveBook.LiveMarkdown.Import do
+  alias LiveBook.Notebook
+  alias LiveBook.LiveMarkdown.MarkdownHelpers
+
+  @doc """
+  Converts the given Markdown document into a notebook data structure.
+
+  Returns the notebook structure and list if informative messages/warnings
+  related to the imported input.
+  """
+  @spec notebook_from_markdown(String.t()) :: {Notebook.t(), list(String.t())}
+  def notebook_from_markdown(markdown) do
+    {_, ast, earmark_messages} = EarmarkParser.as_ast(markdown)
+    earmark_messages = Enum.map(earmark_messages, &earmark_message_to_string/1)
+
+    {ast, rewrite_messages} = rewrite_ast(ast)
+
+    notebook =
+      ast
+      |> group_elements()
+      |> build_notebook()
+
+    {notebook, earmark_messages ++ rewrite_messages}
+  end
+
+  defp earmark_message_to_string({_severity, line_number, message}) do
+    "Line #{line_number}: #{message}"
+  end
+
+  # Does initial pre-processing of the AST, so that it conforms to the expected form.
+  # Returns {altered_ast, messages}.
+  defp rewrite_ast(ast) do
+    {ast, messages1} = rewrite_multiple_primary_headings(ast)
+    {ast, messages2} = move_primary_heading_top(ast)
+
+    {ast, messages1 ++ messages2}
+  end
+
+  # There should be only one h1 tag indicating notebook name,
+  # if there are many we downgrade all headings.
+  # This doesn't apply to documents exported from LiveBook,
+  # but may be the case for an arbitrary markdown file,
+  # so we do our best to preserve the intent.
+  defp rewrite_multiple_primary_headings(ast) do
+    primary_headings = Enum.count(ast, &(tag(&1) == "h1"))
+
+    if primary_headings > 1 do
+      ast = Enum.map(ast, &downgrade_heading/1)
+
+      message =
+        "Downgrading all headings, because #{primary_headings} instances of heading 1 were found"
+
+      {ast, [message]}
+    else
+      {ast, []}
+    end
+  end
+
+  defp downgrade_heading({"h1", attrs, content, meta}), do: {"h2", attrs, content, meta}
+  defp downgrade_heading({"h2", attrs, content, meta}), do: {"h3", attrs, content, meta}
+  defp downgrade_heading({"h3", attrs, content, meta}), do: {"h4", attrs, content, meta}
+  defp downgrade_heading({"h4", attrs, content, meta}), do: {"h5", attrs, content, meta}
+  defp downgrade_heading({"h5", attrs, content, meta}), do: {"h6", attrs, content, meta}
+  defp downgrade_heading({"h6", attrs, content, meta}), do: {"strong", attrs, content, meta}
+  defp downgrade_heading(ast_node), do: ast_node
+
+  # This moves h1 together with any preceding comments to the top.
+  defp move_primary_heading_top(ast) do
+    case Enum.split_while(ast, &(tag(&1) != "h1")) do
+      {_ast, []} ->
+        {ast, []}
+
+      {leading, [heading | rest]} ->
+        {leading, comments} = split_while_right(leading, &(tag(&1) == :comment))
+
+        if leading == [] do
+          {ast, []}
+        else
+          ast = comments ++ [heading] ++ leading ++ rest
+          message = "Moving heading 1 to the top of the notebook"
+          {ast, [message]}
+        end
+    end
+  end
+
+  defp tag(ast_node)
+  defp tag({tag, _, _, _}), do: tag
+  defp tag(_), do: nil
+
+  defp split_while_right(list, fun) do
+    {right_rev, left_rev} = list |> Enum.reverse() |> Enum.split_while(fun)
+    {Enum.reverse(left_rev), Enum.reverse(right_rev)}
+  end
+
+  # Builds a list of classified elements from the AST.
+  defp group_elements(ast, elems \\ [])
+
+  defp group_elements([], elems), do: elems
+
+  defp group_elements([{"h1", _, content, %{}} | ast], elems) do
+    group_elements(ast, [{:notebook_name, content} | elems])
+  end
+
+  defp group_elements([{"h2", _, content, %{}} | ast], elems) do
+    group_elements(ast, [{:section_name, content} | elems])
+  end
+
+  defp group_elements(
+         [{:comment, _, ["live_book:" <> metadata_json], %{comment: true}} | ast],
+         elems
+       ) do
+    group_elements(ast, [{:metadata, metadata_json} | elems])
+  end
+
+  defp group_elements(
+         [{"pre", _, [{"code", [{"class", "elixir"}], [source], %{}}], %{}} | ast],
+         elems
+       ) do
+    group_elements(ast, [{:cell, :elixir, source} | elems])
+  end
+
+  defp group_elements([ast_node | ast], [{:cell, :markdown, md_ast} | rest]) do
+    group_elements(ast, [{:cell, :markdown, [ast_node | md_ast]} | rest])
+  end
+
+  defp group_elements([ast_node | ast], elems) do
+    group_elements(ast, [{:cell, :markdown, [ast_node]} | elems])
+  end
+
+  # Builds a notebook from the list of elements obtained in the previous step.
+  # Note that the list of elements is reversed:
+  # first we group elements by traversing Earmark AST top-down
+  # and then aggregate elements into data strictures going bottom-up.
+  defp build_notebook(elems, cells \\ [], sections \\ [])
+
+  defp build_notebook([{:cell, :elixir, source} | elems], cells, sections) do
+    {metadata, elems} = grab_metadata(elems)
+    cell = %{Notebook.Cell.new(:elixir) | source: source, metadata: metadata}
+    build_notebook(elems, [cell | cells], sections)
+  end
+
+  defp build_notebook([{:cell, :markdown, md_ast} | elems], cells, sections) do
+    {metadata, elems} = grab_metadata(elems)
+    source = md_ast |> Enum.reverse() |> MarkdownHelpers.markdown_from_ast()
+    cell = %{Notebook.Cell.new(:markdown) | source: source, metadata: metadata}
+    build_notebook(elems, [cell | cells], sections)
+  end
+
+  defp build_notebook([{:section_name, content} | elems], cells, sections) do
+    name = MarkdownHelpers.text_from_ast(content)
+    {metadata, elems} = grab_metadata(elems)
+    section = %{Notebook.Section.new() | name: name, cells: cells, metadata: metadata}
+    build_notebook(elems, [], [section | sections])
+  end
+
+  # If there are section-less cells, put them in a default one.
+  defp build_notebook([{:notebook_name, _content} | _] = elems, cells, sections)
+       when cells != [] do
+    section = %{Notebook.Section.new() | cells: cells}
+    build_notebook(elems, [], [section | sections])
+  end
+
+  # If there are section-less cells, put them in a default one.
+  defp build_notebook([] = elems, cells, sections) when cells != [] do
+    section = %{Notebook.Section.new() | cells: cells}
+    build_notebook(elems, [], [section | sections])
+  end
+
+  defp build_notebook([{:notebook_name, content} | elems], [], sections) do
+    name = MarkdownHelpers.text_from_ast(content)
+    {metadata, []} = grab_metadata(elems)
+    %{Notebook.new() | name: name, sections: sections, metadata: metadata}
+  end
+
+  # If there's no explicit notebook heading, use the defaults.
+  defp build_notebook([], [], sections) do
+    %{Notebook.new() | sections: sections}
+  end
+
+  # Takes optional leading metadata JSON object and returns {metadata, rest}.
+  defp grab_metadata([{:metadata, metadata_json} | elems]) do
+    metadata = Jason.decode!(metadata_json)
+    {metadata, elems}
+  end
+
+  defp grab_metadata(elems), do: {%{}, elems}
+end
diff --git a/lib/live_book/live_markdown/markdown_helpers.ex b/lib/live_book/live_markdown/markdown_helpers.ex
new file mode 100644
index 000000000..9852b175c
--- /dev/null
+++ b/lib/live_book/live_markdown/markdown_helpers.ex
@@ -0,0 +1,374 @@
+defmodule LiveBook.LiveMarkdown.MarkdownHelpers do
+  @doc """
+  Reformats the given markdown document.
+  """
+  @spec reformat(String.t()) :: String.t()
+  def reformat(markdown) do
+    markdown
+    |> EarmarkParser.as_ast()
+    |> elem(1)
+    |> markdown_from_ast()
+  end
+
+  @doc """
+  Extracts plain text from the given AST ignoring all the tags.
+  """
+  @spec text_from_ast(EarmarkParser.ast()) :: String.t()
+  def text_from_ast(ast)
+
+  def text_from_ast(ast) when is_list(ast) do
+    ast
+    |> Enum.map(&text_from_ast/1)
+    |> Enum.join("")
+  end
+
+  def text_from_ast(ast) when is_binary(ast), do: ast
+  def text_from_ast({_, _, ast, _}), do: text_from_ast(ast)
+
+  @doc """
+  Renders Markdown string from the given `EarmarkParser` AST.
+  """
+  @spec markdown_from_ast(EarmarkParser.ast()) :: String.t()
+  def markdown_from_ast(ast) do
+    build_md([], ast)
+    |> IO.iodata_to_binary()
+    |> String.trim()
+  end
+
+  defp build_md(iodata, ast)
+
+  defp build_md(iodata, []), do: iodata
+
+  defp build_md(iodata, [string | ast]) when is_binary(string) do
+    string
+    |> append_inline(iodata)
+    |> build_md(ast)
+  end
+
+  defp build_md(iodata, [{tag, attrs, lines, %{verbatim: true}} | ast]) do
+    render_html(tag, attrs, lines)
+    |> append_block(iodata)
+    |> build_md(ast)
+  end
+
+  defp build_md(iodata, [{"em", _, content, %{}} | ast]) do
+    render_emphasis(content)
+    |> append_inline(iodata)
+    |> build_md(ast)
+  end
+
+  defp build_md(iodata, [{"strong", _, content, %{}} | ast]) do
+    render_strong(content)
+    |> append_inline(iodata)
+    |> build_md(ast)
+  end
+
+  defp build_md(iodata, [{"del", _, content, %{}} | ast]) do
+    render_strikethrough(content)
+    |> append_inline(iodata)
+    |> build_md(ast)
+  end
+
+  defp build_md(iodata, [{"code", _, content, %{}} | ast]) do
+    render_code_inline(content)
+    |> append_inline(iodata)
+    |> build_md(ast)
+  end
+
+  defp build_md(iodata, [{"a", attrs, content, %{}} | ast]) do
+    render_link(content, attrs)
+    |> append_inline(iodata)
+    |> build_md(ast)
+  end
+
+  defp build_md(iodata, [{"img", attrs, [], %{}} | ast]) do
+    render_image(attrs)
+    |> append_inline(iodata)
+    |> build_md(ast)
+  end
+
+  defp build_md(iodata, [{:comment, _, lines, %{comment: true}} | ast]) do
+    render_comment(lines)
+    |> append_inline(iodata)
+    |> build_md(ast)
+  end
+
+  defp build_md(iodata, [{"hr", attrs, [], %{}} | ast]) do
+    render_ruler(attrs)
+    |> append_block(iodata)
+    |> build_md(ast)
+  end
+
+  defp build_md(iodata, [{"p", _, content, %{}} | ast]) do
+    render_paragraph(content)
+    |> append_block(iodata)
+    |> build_md(ast)
+  end
+
+  defp build_md(iodata, [{"h" <> n, _, content, %{}} | ast])
+       when n in ["1", "2", "3", "4", "5", "6"] do
+    n = String.to_integer(n)
+
+    render_heading(n, content)
+    |> append_block(iodata)
+    |> build_md(ast)
+  end
+
+  defp build_md(iodata, [{"pre", _, [{"code", attrs, [content], %{}}], %{}} | ast]) do
+    render_code_block(content, attrs)
+    |> append_block(iodata)
+    |> build_md(ast)
+  end
+
+  defp build_md(iodata, [{"blockquote", [], content, %{}} | ast]) do
+    render_blockquote(content)
+    |> append_block(iodata)
+    |> build_md(ast)
+  end
+
+  defp build_md(iodata, [{"table", _, content, %{}} | ast]) do
+    render_table(content)
+    |> append_block(iodata)
+    |> build_md(ast)
+  end
+
+  defp build_md(iodata, [{"ul", _, content, %{}} | ast]) do
+    render_unordered_list(content)
+    |> append_block(iodata)
+    |> build_md(ast)
+  end
+
+  defp build_md(iodata, [{"ol", _, content, %{}} | ast]) do
+    render_ordered_list(content)
+    |> append_block(iodata)
+    |> build_md(ast)
+  end
+
+  defp append_inline(md, iodata), do: [iodata, md]
+  defp append_block(md, iodata), do: [iodata, "\n", md, "\n"]
+
+  # Renderers
+
+  # https://www.w3.org/TR/2011/WD-html-markup-20110113/syntax.html#void-element
+  @void_elements ~W(area base br col command embed hr img input keygen link meta param source track wbr)
+
+  defp render_html(tag, attrs, []) when tag in @void_elements do
+    "<#{tag} #{attrs_to_string(attrs)} />"
+  end
+
+  defp render_html(tag, attrs, lines) do
+    inner = Enum.join(lines, "\n")
+    "<#{tag} #{attrs_to_string(attrs)}>\n#{inner}\n</#{tag}>"
+  end
+
+  defp render_emphasis(content) do
+    inner = markdown_from_ast(content)
+    "*#{inner}*"
+  end
+
+  defp render_strong(content) do
+    inner = markdown_from_ast(content)
+    "**#{inner}**"
+  end
+
+  defp render_strikethrough(content) do
+    inner = markdown_from_ast(content)
+    "~~#{inner}~~"
+  end
+
+  defp render_code_inline(content) do
+    inner = markdown_from_ast(content)
+    "`#{inner}`"
+  end
+
+  defp render_link(content, attrs) do
+    caption = markdown_from_ast(content)
+    href = get_attr(attrs, "href", "")
+    "[#{caption}](#{href})"
+  end
+
+  defp render_image(attrs) do
+    alt = get_attr(attrs, "alt", "")
+    src = get_attr(attrs, "src", "")
+    title = get_attr(attrs, "title", "")
+
+    if title == "" do
+      "![#{alt}](#{src})"
+    else
+      ~s/![#{alt}](#{src} "#{title}")/
+    end
+  end
+
+  defp render_comment([line]) do
+    line = String.trim(line)
+    "<!-- #{line} -->"
+  end
+
+  defp render_comment(lines) do
+    lines =
+      lines
+      |> Enum.drop_while(&blank?/1)
+      |> Enum.reverse()
+      |> Enum.drop_while(&blank?/1)
+      |> Enum.reverse()
+
+    Enum.join(["<!--" | lines] ++ ["-->"], "\n")
+  end
+
+  defp render_ruler(attrs) do
+    class = get_attr(attrs, "class", "thin")
+
+    case class do
+      "thin" -> "---"
+      "medium" -> "___"
+      "thick" -> "***"
+    end
+  end
+
+  defp render_paragraph(content), do: markdown_from_ast(content)
+
+  defp render_heading(n, content) do
+    title = markdown_from_ast(content)
+    String.duplicate("#", n) <> " " <> title
+  end
+
+  defp render_code_block(content, attrs) do
+    language = get_attr(attrs, "class", "")
+    "```#{language}\n#{content}\n```"
+  end
+
+  defp render_blockquote(content) do
+    inner = markdown_from_ast(content)
+
+    inner
+    |> String.split("\n")
+    |> Enum.map(&("> " <> &1))
+    |> Enum.join("\n")
+  end
+
+  defp render_table([{"thead", _, [head_row], %{}}, {"tbody", _, body_rows, %{}}]) do
+    alignments = alignments_from_row(head_row)
+    cell_grid = cell_grid_from_rows([head_row | body_rows])
+    column_widths = max_length_per_column(cell_grid)
+    [head_cells | body_cell_grid] = Enum.map(cell_grid, &pad_whitespace(&1, column_widths))
+    separator_cells = build_separator_cells(alignments, column_widths)
+    cell_grid_to_md_table([head_cells, separator_cells | body_cell_grid])
+  end
+
+  defp render_table([{"tbody", _, body_rows, %{}}]) do
+    cell_grid = cell_grid_from_rows(body_rows)
+    column_widths = max_length_per_column(cell_grid)
+    cell_grid = Enum.map(cell_grid, &pad_whitespace(&1, column_widths))
+    cell_grid_to_md_table(cell_grid)
+  end
+
+  defp cell_grid_from_rows(rows) do
+    Enum.map(rows, fn {"tr", _, columns, %{}} ->
+      Enum.map(columns, fn {tag, _, content, %{}} when tag in ["th", "td"] ->
+        markdown_from_ast(content)
+      end)
+    end)
+  end
+
+  defp alignments_from_row({"tr", _, columns, %{}}) do
+    Enum.map(columns, fn {tag, attrs, _, %{}} when tag in ["th", "td"] ->
+      style = get_attr(attrs, "style", nil)
+
+      case style do
+        "text-align: left;" -> :left
+        "text-align: center;" -> :center
+        "text-align: right;" -> :right
+      end
+    end)
+  end
+
+  defp build_separator_cells(alignments, widths) do
+    alignments
+    |> Enum.zip(widths)
+    |> Enum.map(fn
+      {:left, length} -> String.duplicate("-", length)
+      {:center, length} -> ":" <> String.duplicate("-", length - 2) <> ":"
+      {:right, length} -> String.duplicate("-", length - 1) <> ":"
+    end)
+  end
+
+  defp max_length_per_column(cell_grid) do
+    cell_grid
+    |> List.zip()
+    |> Enum.map(&Tuple.to_list/1)
+    |> Enum.map(fn cells ->
+      cells
+      |> Enum.map(&String.length/1)
+      |> Enum.max()
+    end)
+  end
+
+  defp pad_whitespace(cells, widths) do
+    cells
+    |> Enum.zip(widths)
+    |> Enum.map(fn {cell, width} ->
+      String.pad_trailing(cell, width, " ")
+    end)
+  end
+
+  defp cell_grid_to_md_table(cell_grid) do
+    cell_grid
+    |> Enum.map(fn cells ->
+      "| " <> Enum.join(cells, " | ") <> " |"
+    end)
+    |> Enum.join("\n")
+  end
+
+  defp render_unordered_list(content) do
+    marker_fun = fn _index -> "* " end
+    render_list(content, marker_fun, "  ")
+  end
+
+  defp render_ordered_list(content) do
+    marker_fun = fn index -> "#{index + 1}. " end
+    render_list(content, marker_fun, "   ")
+  end
+
+  defp render_list(items, marker_fun, indent) do
+    spaced? = spaced_list_items?(items)
+    item_separator = if(spaced?, do: "\n\n", else: "\n")
+
+    items
+    |> Enum.map(fn {"li", _, content, %{}} -> markdown_from_ast(content) end)
+    |> Enum.with_index()
+    |> Enum.map(fn {inner, index} ->
+      [first_line | lines] = String.split(inner, "\n")
+
+      first_line = marker_fun.(index) <> first_line
+
+      lines =
+        Enum.map(lines, fn
+          "" -> ""
+          line -> indent <> line
+        end)
+
+      Enum.join([first_line | lines], "\n")
+    end)
+    |> Enum.join(item_separator)
+  end
+
+  defp spaced_list_items?([{"li", _, [{"p", _, _content, %{}} | _], %{}} | _items]), do: true
+  defp spaced_list_items?([_ | items]), do: spaced_list_items?(items)
+  defp spaced_list_items?([]), do: false
+
+  # Helpers
+
+  defp get_attr(attrs, key, default) do
+    Enum.find_value(attrs, default, fn {attr_key, attr_value} ->
+      attr_key == key && attr_value
+    end)
+  end
+
+  defp attrs_to_string(attrs) do
+    attrs
+    |> Enum.map(fn {key, value} -> ~s/#{key}="#{value}"/ end)
+    |> Enum.join(" ")
+  end
+
+  defp blank?(string), do: String.trim(string) == ""
+end
diff --git a/lib/live_book/notebook.ex b/lib/live_book/notebook.ex
index ef95f9ef2..ea76b76ba 100644
--- a/lib/live_book/notebook.ex
+++ b/lib/live_book/notebook.ex
@@ -19,7 +19,7 @@ defmodule LiveBook.Notebook do
           name: String.t(),
           version: String.t(),
           sections: list(Section.t()),
-          metadata: %{atom() => term()}
+          metadata: %{String.t() => term()}
         }
 
   @version "1.0"
diff --git a/lib/live_book/notebook/cell.ex b/lib/live_book/notebook/cell.ex
index bc540333c..50bfdbc80 100644
--- a/lib/live_book/notebook/cell.ex
+++ b/lib/live_book/notebook/cell.ex
@@ -19,7 +19,7 @@ defmodule LiveBook.Notebook.Cell do
           type: type(),
           source: String.t(),
           outputs: list(),
-          metadata: %{atom() => term()}
+          metadata: %{String.t() => term()}
         }
 
   @doc """
diff --git a/lib/live_book/notebook/section.ex b/lib/live_book/notebook/section.ex
index f80b59d73..84afe8de7 100644
--- a/lib/live_book/notebook/section.ex
+++ b/lib/live_book/notebook/section.ex
@@ -17,7 +17,7 @@ defmodule LiveBook.Notebook.Section do
           id: id(),
           name: String.t(),
           cells: list(Cell.t()),
-          metadata: %{atom() => term()}
+          metadata: %{String.t() => term()}
         }
 
   @doc """
diff --git a/lib/live_book_web/live/cell_component.ex b/lib/live_book_web/live/cell_component.ex
index 04785d517..5f3e50e1b 100644
--- a/lib/live_book_web/live/cell_component.ex
+++ b/lib/live_book_web/live/cell_component.ex
@@ -3,7 +3,7 @@ defmodule LiveBookWeb.CellComponent do
 
   def render(assigns) do
     ~L"""
-    <div class="flex flex-col relative mr-10 border-l-4 pl-4 -ml-4 border-blue-100 border-opacity-0 hover:border-opacity-100 <%= if @focused, do: "border-blue-300 border-opacity-100"%>"
+    <div class="cell flex flex-col relative mr-10 border-l-4 pl-4 -ml-4 border-blue-100 border-opacity-0 hover:border-opacity-100 <%= if @focused, do: "border-blue-300 border-opacity-100"%>"
       id="cell-<%= @cell.id %>"
       phx-hook="Cell"
       data-cell-id="<%= @cell.id %>"
diff --git a/mix.exs b/mix.exs
index 8e5ebab20..a421ff360 100644
--- a/mix.exs
+++ b/mix.exs
@@ -41,7 +41,8 @@ defmodule LiveBook.MixProject do
       {:telemetry_metrics, "~> 0.4"},
       {:telemetry_poller, "~> 0.4"},
       {:jason, "~> 1.0"},
-      {:plug_cowboy, "~> 2.0"}
+      {:plug_cowboy, "~> 2.0"},
+      {:earmark_parser, "~> 1.4"}
     ]
   end
 
diff --git a/mix.lock b/mix.lock
index 31d2a80ec..b30517609 100644
--- a/mix.lock
+++ b/mix.lock
@@ -2,15 +2,16 @@
   "cowboy": {:hex, :cowboy, "2.8.0", "f3dc62e35797ecd9ac1b50db74611193c29815401e53bac9a5c0577bd7bc667d", [:rebar3], [{:cowlib, "~> 2.9.1", [hex: :cowlib, repo: "hexpm", optional: false]}, {:ranch, "~> 1.7.1", [hex: :ranch, repo: "hexpm", optional: false]}], "hexpm", "4643e4fba74ac96d4d152c75803de6fad0b3fa5df354c71afdd6cbeeb15fac8a"},
   "cowboy_telemetry": {:hex, :cowboy_telemetry, "0.3.1", "ebd1a1d7aff97f27c66654e78ece187abdc646992714164380d8a041eda16754", [:rebar3], [{:cowboy, "~> 2.7", [hex: :cowboy, repo: "hexpm", optional: false]}, {:telemetry, "~> 0.4", [hex: :telemetry, repo: "hexpm", optional: false]}], "hexpm", "3a6efd3366130eab84ca372cbd4a7d3c3a97bdfcfb4911233b035d117063f0af"},
   "cowlib": {:hex, :cowlib, "2.9.1", "61a6c7c50cf07fdd24b2f45b89500bb93b6686579b069a89f88cb211e1125c78", [:rebar3], [], "hexpm", "e4175dc240a70d996156160891e1c62238ede1729e45740bdd38064dad476170"},
+  "earmark_parser": {:hex, :earmark_parser, "1.4.12", "b245e875ec0a311a342320da0551da407d9d2b65d98f7a9597ae078615af3449", [:mix], [], "hexpm", "711e2cc4d64abb7d566d43f54b78f7dc129308a63bc103fbd88550d2174b3160"},
   "file_system": {:hex, :file_system, "0.2.10", "fb082005a9cd1711c05b5248710f8826b02d7d1784e7c3451f9c1231d4fc162d", [:mix], [], "hexpm", "41195edbfb562a593726eda3b3e8b103a309b733ad25f3d642ba49696bf715dc"},
-  "floki": {:hex, :floki, "0.29.0", "b1710d8c93a2f860dc2d7adc390dd808dc2fb8f78ee562304457b75f4c640881", [:mix], [{:html_entities, "~> 0.5.0", [hex: :html_entities, repo: "hexpm", optional: false]}], "hexpm", "008585ce64b9f74c07d32958ec9866f4b8a124bf4da1e2941b28e41384edaaad"},
+  "floki": {:hex, :floki, "0.30.0", "22ebbe681a5d3777cdd830ca091b1b806d33c3449c26312eadca7f7be685c0c8", [:mix], [{:html_entities, "~> 0.5.0", [hex: :html_entities, repo: "hexpm", optional: false]}], "hexpm", "a9e128a4ca9bb71f11affa315b6768a9ad326d5996ff1e92acf1d7a01a10076a"},
   "html_entities": {:hex, :html_entities, "0.5.1", "1c9715058b42c35a2ab65edc5b36d0ea66dd083767bef6e3edb57870ef556549", [:mix], [], "hexpm", "30efab070904eb897ff05cd52fa61c1025d7f8ef3a9ca250bc4e6513d16c32de"},
   "jason": {:hex, :jason, "1.2.2", "ba43e3f2709fd1aa1dce90aaabfd039d000469c05c56f0b8e31978e03fa39052", [:mix], [{:decimal, "~> 1.0 or ~> 2.0", [hex: :decimal, repo: "hexpm", optional: true]}], "hexpm", "18a228f5f0058ee183f29f9eae0805c6e59d61c3b006760668d8d18ff0d12179"},
   "mime": {:hex, :mime, "1.5.0", "203ef35ef3389aae6d361918bf3f952fa17a09e8e43b5aa592b93eba05d0fb8d", [:mix], [], "hexpm", "55a94c0f552249fc1a3dd9cd2d3ab9de9d3c89b559c2bd01121f824834f24746"},
   "phoenix": {:hex, :phoenix, "1.5.7", "2923bb3af924f184459fe4fa4b100bd25fa6468e69b2803dfae82698269aa5e0", [:mix], [{:jason, "~> 1.0", [hex: :jason, repo: "hexpm", optional: true]}, {:phoenix_html, "~> 2.13", [hex: :phoenix_html, repo: "hexpm", optional: true]}, {:phoenix_pubsub, "~> 2.0", [hex: :phoenix_pubsub, repo: "hexpm", optional: false]}, {:plug, "~> 1.10", [hex: :plug, repo: "hexpm", optional: false]}, {:plug_cowboy, "~> 1.0 or ~> 2.2", [hex: :plug_cowboy, repo: "hexpm", optional: true]}, {:plug_crypto, "~> 1.1.2 or ~> 1.2", [hex: :plug_crypto, repo: "hexpm", optional: false]}, {:telemetry, "~> 0.4", [hex: :telemetry, repo: "hexpm", optional: false]}], "hexpm", "774cd64417c5a3788414fdbb2be2eb9bcd0c048d9e6ad11a0c1fd67b7c0d0978"},
   "phoenix_html": {:hex, :phoenix_html, "2.14.3", "51f720d0d543e4e157ff06b65de38e13303d5778a7919bcc696599e5934271b8", [:mix], [{:plug, "~> 1.5", [hex: :plug, repo: "hexpm", optional: false]}], "hexpm", "efd697a7fff35a13eeeb6b43db884705cba353a1a41d127d118fda5f90c8e80f"},
   "phoenix_live_reload": {:hex, :phoenix_live_reload, "1.3.0", "f35f61c3f959c9a01b36defaa1f0624edd55b87e236b606664a556d6f72fd2e7", [:mix], [{:file_system, "~> 0.2.1 or ~> 0.3", [hex: :file_system, repo: "hexpm", optional: false]}, {:phoenix, "~> 1.4", [hex: :phoenix, repo: "hexpm", optional: false]}], "hexpm", "02c1007ae393f2b76ec61c1a869b1e617179877984678babde131d716f95b582"},
-  "phoenix_live_view": {:hex, :phoenix_live_view, "0.15.3", "70c7917e5c421e32d1a1c8ddf8123378bb741748cd8091eb9d557fb4be92a94f", [:mix], [{:jason, "~> 1.0", [hex: :jason, repo: "hexpm", optional: true]}, {:phoenix, "~> 1.5.7", [hex: :phoenix, repo: "hexpm", optional: false]}, {:phoenix_html, "~> 2.14", [hex: :phoenix_html, repo: "hexpm", optional: false]}, {:telemetry, "~> 0.4.2 or ~> 0.5", [hex: :telemetry, repo: "hexpm", optional: false]}], "hexpm", "cabcfb6738419a08600009219a5f0d861de97507fc1232121e1d5221aba849bd"},
+  "phoenix_live_view": {:hex, :phoenix_live_view, "0.15.4", "86908dc9603cc81c07e84725ee42349b5325cb250c9c20d3533856ff18dbb7dc", [:mix], [{:jason, "~> 1.0", [hex: :jason, repo: "hexpm", optional: true]}, {:phoenix, "~> 1.5.7", [hex: :phoenix, repo: "hexpm", optional: false]}, {:phoenix_html, "~> 2.14", [hex: :phoenix_html, repo: "hexpm", optional: false]}, {:telemetry, "~> 0.4.2 or ~> 0.5", [hex: :telemetry, repo: "hexpm", optional: false]}], "hexpm", "35d78f3c35fe10a995dca5f4ab50165b7a90cbe02e23de245381558f821e9462"},
   "phoenix_pubsub": {:hex, :phoenix_pubsub, "2.0.0", "a1ae76717bb168cdeb10ec9d92d1480fec99e3080f011402c0a2d68d47395ffb", [:mix], [], "hexpm", "c52d948c4f261577b9c6fa804be91884b381a7f8f18450c5045975435350f771"},
   "plug": {:hex, :plug, "1.11.0", "f17217525597628298998bc3baed9f8ea1fa3f1160aa9871aee6df47a6e4d38e", [:mix], [{:mime, "~> 1.0", [hex: :mime, repo: "hexpm", optional: false]}, {:plug_crypto, "~> 1.1.1 or ~> 1.2", [hex: :plug_crypto, repo: "hexpm", optional: false]}, {:telemetry, "~> 0.4", [hex: :telemetry, repo: "hexpm", optional: false]}], "hexpm", "2d9c633f0499f9dc5c2fd069161af4e2e7756890b81adcbb2ceaa074e8308876"},
   "plug_cowboy": {:hex, :plug_cowboy, "2.4.1", "779ba386c0915027f22e14a48919a9545714f849505fa15af2631a0d298abf0f", [:mix], [{:cowboy, "~> 2.7", [hex: :cowboy, repo: "hexpm", optional: false]}, {:cowboy_telemetry, "~> 0.3", [hex: :cowboy_telemetry, repo: "hexpm", optional: false]}, {:plug, "~> 1.7", [hex: :plug, repo: "hexpm", optional: false]}, {:telemetry, "~> 0.4", [hex: :telemetry, repo: "hexpm", optional: false]}], "hexpm", "d72113b6dff7b37a7d9b2a5b68892808e3a9a752f2bf7e503240945385b70507"},
diff --git a/test/live_book/live_markdown/export_test.exs b/test/live_book/live_markdown/export_test.exs
new file mode 100644
index 000000000..927837271
--- /dev/null
+++ b/test/live_book/live_markdown/export_test.exs
@@ -0,0 +1,224 @@
+defmodule LiveBook.LiveMarkdown.ExportTest do
+  use ExUnit.Case, async: true
+
+  alias LiveBook.LiveMarkdown.Export
+  alias LiveBook.Notebook
+
+  test "acceptance" do
+    notebook = %{
+      Notebook.new()
+      | name: "My Notebook",
+        metadata: %{"author" => "Sherlock Holmes"},
+        sections: [
+          %{
+            Notebook.Section.new()
+            | name: "Section 1",
+              metadata: %{"created_at" => "2021-02-15"},
+              cells: [
+                %{
+                  Notebook.Cell.new(:markdown)
+                  | metadata: %{"updated_at" => "2021-02-15"},
+                    source: """
+                    Make sure to install:
+
+                    * Erlang
+                    * Elixir
+                    * PostgreSQL\
+                    """
+                },
+                %{
+                  Notebook.Cell.new(:elixir)
+                  | metadata: %{"readonly" => true},
+                    source: """
+                    Enum.to_list(1..10)\
+                    """
+                },
+                %{
+                  Notebook.Cell.new(:markdown)
+                  | metadata: %{},
+                    source: """
+                    This is it for this section.\
+                    """
+                }
+              ]
+          },
+          %{
+            Notebook.Section.new()
+            | name: "Section 2",
+              metadata: %{},
+              cells: [
+                %{
+                  Notebook.Cell.new(:elixir)
+                  | metadata: %{},
+                    source: """
+                    # More Elixir code\
+                    """
+                }
+              ]
+          }
+        ]
+    }
+
+    expected_document = """
+    <!--live_book:{"author":"Sherlock Holmes"}-->
+    # My Notebook
+
+    <!--live_book:{"created_at":"2021-02-15"}-->
+    ## Section 1
+
+    <!--live_book:{"updated_at":"2021-02-15"}-->
+    Make sure to install:
+
+    * Erlang
+    * Elixir
+    * PostgreSQL
+
+    <!--live_book:{"readonly":true}-->
+    ```elixir
+    Enum.to_list(1..10)
+    ```
+
+    This is it for this section.
+
+    ## Section 2
+
+    ```elixir
+    # More Elixir code
+    ```
+    """
+
+    document = Export.notebook_to_markdown(notebook)
+
+    assert expected_document == document
+  end
+
+  test "reformats markdown cells" do
+    notebook = %{
+      Notebook.new()
+      | name: "My Notebook",
+        metadata: %{},
+        sections: [
+          %{
+            Notebook.Section.new()
+            | name: "Section 1",
+              metadata: %{},
+              cells: [
+                %{
+                  Notebook.Cell.new(:markdown)
+                  | metadata: %{},
+                    source: """
+                    |State|Abbrev|Capital|
+                    | --: | :-: | --- |
+                    | Texas | TX | Austin |
+                    | Maine | ME | Augusta |
+                    """
+                }
+              ]
+          }
+        ]
+    }
+
+    expected_document = """
+    # My Notebook
+
+    ## Section 1
+
+    | State | Abbrev | Capital |
+    | ----: | :----: | ------- |
+    | Texas | TX     | Austin  |
+    | Maine | ME     | Augusta |
+    """
+
+    document = Export.notebook_to_markdown(notebook)
+
+    assert expected_document == document
+  end
+
+  test "drops heading 1 and 2 in markdown cells" do
+    notebook = %{
+      Notebook.new()
+      | name: "My Notebook",
+        metadata: %{},
+        sections: [
+          %{
+            Notebook.Section.new()
+            | name: "Section 1",
+              metadata: %{},
+              cells: [
+                %{
+                  Notebook.Cell.new(:markdown)
+                  | metadata: %{},
+                    source: """
+                    # Heading 1
+
+                    ## Heading 2
+
+                    ### Heading 3
+                    """
+                }
+              ]
+          }
+        ]
+    }
+
+    expected_document = """
+    # My Notebook
+
+    ## Section 1
+
+    ### Heading 3
+    """
+
+    document = Export.notebook_to_markdown(notebook)
+
+    assert expected_document == document
+  end
+
+  test "keeps non-elixir code snippets" do
+    notebook = %{
+      Notebook.new()
+      | name: "My Notebook",
+        metadata: %{},
+        sections: [
+          %{
+            Notebook.Section.new()
+            | name: "Section 1",
+              metadata: %{},
+              cells: [
+                %{
+                  Notebook.Cell.new(:markdown)
+                  | metadata: %{},
+                    source: """
+                    ```shell
+                    mix deps.get
+                    ```
+
+                    ```erlang
+                    spawn_link(fun() -> io:format("Hiya") end).
+                    ```
+                    """
+                }
+              ]
+          }
+        ]
+    }
+
+    expected_document = """
+    # My Notebook
+
+    ## Section 1
+
+    ```shell
+    mix deps.get
+    ```
+
+    ```erlang
+    spawn_link(fun() -> io:format("Hiya") end).
+    ```
+    """
+
+    document = Export.notebook_to_markdown(notebook)
+
+    assert expected_document == document
+  end
+end
diff --git a/test/live_book/live_markdown/import_test.exs b/test/live_book/live_markdown/import_test.exs
new file mode 100644
index 000000000..08fe02477
--- /dev/null
+++ b/test/live_book/live_markdown/import_test.exs
@@ -0,0 +1,378 @@
+defmodule LiveBook.LiveMarkdown.ImportTest do
+  use ExUnit.Case, async: true
+
+  alias LiveBook.LiveMarkdown.Import
+  alias LiveBook.Notebook
+
+  test "acceptance" do
+    markdown = """
+    <!--live_book:{"author":"Sherlock Holmes"}-->
+    # My Notebook
+
+    <!--live_book:{"created_at":"2021-02-15"}-->
+    ## Section 1
+
+    <!--live_book:{"updated_at":"2021-02-15"}-->
+    Make sure to install:
+
+    * Erlang
+    * Elixir
+    * PostgreSQL
+
+    <!--live_book:{"readonly":true}-->
+    ```elixir
+    Enum.to_list(1..10)
+    ```
+
+    This is it for this section.
+
+    ## Section 2
+
+    ```elixir
+    # More Elixir code
+    ```
+    """
+
+    {notebook, []} = Import.notebook_from_markdown(markdown)
+
+    # Match only on the relevant fields as some may be generated (ids).
+
+    assert %Notebook{
+             name: "My Notebook",
+             metadata: %{"author" => "Sherlock Holmes"},
+             sections: [
+               %Notebook.Section{
+                 name: "Section 1",
+                 metadata: %{"created_at" => "2021-02-15"},
+                 cells: [
+                   %Notebook.Cell{
+                     type: :markdown,
+                     metadata: %{"updated_at" => "2021-02-15"},
+                     source: """
+                     Make sure to install:
+
+                     * Erlang
+                     * Elixir
+                     * PostgreSQL\
+                     """
+                   },
+                   %Notebook.Cell{
+                     type: :elixir,
+                     metadata: %{"readonly" => true},
+                     source: """
+                     Enum.to_list(1..10)\
+                     """
+                   },
+                   %Notebook.Cell{
+                     type: :markdown,
+                     metadata: %{},
+                     source: """
+                     This is it for this section.\
+                     """
+                   }
+                 ]
+               },
+               %Notebook.Section{
+                 name: "Section 2",
+                 metadata: %{},
+                 cells: [
+                   %Notebook.Cell{
+                     metadata: %{},
+                     source: """
+                     # More Elixir code\
+                     """
+                   }
+                 ]
+               }
+             ]
+           } = notebook
+  end
+
+  test "reformats markdown cells" do
+    markdown = """
+    # My Notebook
+
+    ## Section 1
+
+    |State|Abbrev|Capital|
+    | --: | :-: | --- |
+    | Texas | TX | Austin |
+    | Maine | ME | Augusta |
+    """
+
+    {notebook, []} = Import.notebook_from_markdown(markdown)
+
+    assert %Notebook{
+             name: "My Notebook",
+             sections: [
+               %Notebook.Section{
+                 name: "Section 1",
+                 metadata: %{},
+                 cells: [
+                   %Notebook.Cell{
+                     type: :markdown,
+                     metadata: %{},
+                     source: """
+                     | State | Abbrev | Capital |
+                     | ----: | :----: | ------- |
+                     | Texas | TX     | Austin  |
+                     | Maine | ME     | Augusta |\
+                     """
+                   }
+                 ]
+               }
+             ]
+           } = notebook
+  end
+
+  test "uses default name if there is no primary heading" do
+    markdown = """
+    ## Section 1
+
+    Some markdown.
+    """
+
+    {notebook, []} = Import.notebook_from_markdown(markdown)
+
+    assert %Notebook{
+             name: "Untitled notebook",
+             sections: [
+               %Notebook.Section{
+                 name: "Section 1"
+               }
+             ]
+           } = notebook
+  end
+
+  test "given multiple primary heading, downgrades all headings" do
+    markdown = """
+    # Probably section 1
+
+    ## Heading
+
+    Some markdown.
+
+    # Probably section 2
+
+    ###### Tiny heading
+    """
+
+    {notebook, messages} = Import.notebook_from_markdown(markdown)
+
+    assert %Notebook{
+             name: "Untitled notebook",
+             sections: [
+               %Notebook.Section{
+                 name: "Probably section 1",
+                 cells: [
+                   %Notebook.Cell{
+                     type: :markdown,
+                     metadata: %{},
+                     source: """
+                     ### Heading
+
+                     Some markdown.\
+                     """
+                   }
+                 ]
+               },
+               %Notebook.Section{
+                 name: "Probably section 2",
+                 cells: [
+                   %Notebook.Cell{
+                     type: :markdown,
+                     metadata: %{},
+                     source: """
+                     **Tiny heading**\
+                     """
+                   }
+                 ]
+               }
+             ]
+           } = notebook
+
+    assert ["Downgrading all headings, because 2 instances of heading 1 were found"] == messages
+  end
+
+  test "ignores markdown modifiers in notebok/section names" do
+    markdown = """
+    # My *Notebook*
+
+    ## [Section 1](https://example.com)
+    """
+
+    {notebook, []} = Import.notebook_from_markdown(markdown)
+
+    assert %Notebook{
+             name: "My Notebook",
+             sections: [
+               %Notebook.Section{
+                 name: "Section 1"
+               }
+             ]
+           } = notebook
+  end
+
+  test "adds a default section if there is some section-less content" do
+    markdown = """
+    # My Notebook
+
+    Some markdown.
+
+    ## Actual section
+    """
+
+    {notebook, []} = Import.notebook_from_markdown(markdown)
+
+    assert %Notebook{
+             name: "My Notebook",
+             sections: [
+               %Notebook.Section{
+                 name: "Section",
+                 cells: [
+                   %Notebook.Cell{
+                     type: :markdown,
+                     source: """
+                     Some markdown.\
+                     """
+                   }
+                 ]
+               },
+               %Notebook.Section{
+                 name: "Actual section"
+               }
+             ]
+           } = notebook
+  end
+
+  test "uses defaults if there are no headings" do
+    markdown = """
+    ```elixir
+    Enum.to_list(1..10)
+    ```
+    """
+
+    {notebook, []} = Import.notebook_from_markdown(markdown)
+
+    assert %Notebook{
+             name: "Untitled notebook",
+             sections: [
+               %Notebook.Section{
+                 name: "Section",
+                 cells: [
+                   %Notebook.Cell{
+                     type: :elixir,
+                     source: """
+                     Enum.to_list(1..10)\
+                     """
+                   }
+                 ]
+               }
+             ]
+           } = notebook
+  end
+
+  test "moves the primary heading and preceding comments to the top" do
+    markdown = """
+    Cool notebook.
+
+    <!--live_book:{"author":"Sherlock Holmes"}-->
+    # My Notebook
+
+    Some markdown.
+    """
+
+    {notebook, messages} = Import.notebook_from_markdown(markdown)
+
+    assert %Notebook{
+             name: "My Notebook",
+             metadata: %{"author" => "Sherlock Holmes"},
+             sections: [
+               %Notebook.Section{
+                 name: "Section",
+                 cells: [
+                   %Notebook.Cell{
+                     type: :markdown,
+                     source: """
+                     Cool notebook.
+
+                     Some markdown.\
+                     """
+                   }
+                 ]
+               }
+             ]
+           } = notebook
+
+    assert ["Moving heading 1 to the top of the notebook"] == messages
+  end
+
+  test "includes parsing warnings in the returned message list" do
+    markdown = """
+    # My notebook
+
+    `
+
+    Some markdown.
+    """
+
+    {_notebook, messages} = Import.notebook_from_markdown(markdown)
+
+    assert ["Line 3: Closing unclosed backquotes ` at end of input"] == messages
+  end
+
+  test "imports non-elixir code snippets as part of markdown cells" do
+    markdown = """
+    # My Notebook
+
+    ## Section 1
+
+    ```shell
+    mix deps.get
+    ```
+
+    ```elixir
+    Enum.to_list(1..10)
+    ```
+
+    ```erlang
+    spawn_link(fun() -> io:format("Hiya") end).
+    ```
+    """
+
+    {notebook, []} = Import.notebook_from_markdown(markdown)
+
+    assert %Notebook{
+             name: "My Notebook",
+             sections: [
+               %Notebook.Section{
+                 name: "Section 1",
+                 cells: [
+                   %Notebook.Cell{
+                     type: :markdown,
+                     source: """
+                     ```shell
+                     mix deps.get
+                     ```\
+                     """
+                   },
+                   %Notebook.Cell{
+                     type: :elixir,
+                     source: """
+                     Enum.to_list(1..10)\
+                     """
+                   },
+                   %Notebook.Cell{
+                     type: :markdown,
+                     source: """
+                     ```erlang
+                     spawn_link(fun() -> io:format("Hiya") end).
+                     ```\
+                     """
+                   }
+                 ]
+               }
+             ]
+           } = notebook
+  end
+end
diff --git a/test/live_book/live_markdown/markdown_helpers_test.exs b/test/live_book/live_markdown/markdown_helpers_test.exs
new file mode 100644
index 000000000..feee417c5
--- /dev/null
+++ b/test/live_book/live_markdown/markdown_helpers_test.exs
@@ -0,0 +1,281 @@
+defmodule LiveBook.LiveMarkdown.MarkdownHelpersTest do
+  use ExUnit.Case, async: true
+
+  alias LiveBook.LiveMarkdown.MarkdownHelpers
+
+  describe "markdown_from_ast/1" do
+    test "emphasis" do
+      markdown = "The Game, *Mrs Hudson*, is on!"
+      assert markdown == reformat(markdown)
+    end
+
+    test "bold" do
+      markdown = "The Game, **Mrs Hudson**, is on!"
+      assert markdown == reformat(markdown)
+    end
+
+    test "strikethrough" do
+      markdown = "The Game, ~~Mrs Hudson~~, is on!"
+      assert markdown == reformat(markdown)
+    end
+
+    test "inline code" do
+      markdown = "The Game, `Mrs Hudson`, is on!"
+      assert markdown == reformat(markdown)
+    end
+
+    test "combined" do
+      markdown = "The Game, ~~***`Mrs Hudson`***~~, is on!"
+      assert markdown == reformat(markdown)
+    end
+
+    test "link" do
+      markdown = "The Game, [Mrs Hudson](https://youtu.be/M-KqaO1oH2E), is on!"
+      assert markdown == reformat(markdown)
+    end
+
+    test "basic image" do
+      markdown = "The Game, ![Mrs Hudson](https://example.com), is on!"
+      assert markdown == reformat(markdown)
+    end
+
+    test "image with title" do
+      markdown = "The Game, ![Mrs Hudson](https://example.com \"Title\"), is on!"
+      assert markdown == reformat(markdown)
+    end
+
+    test "img tag with additional attributes is kept as a tag" do
+      markdown = ~s{<img src="https://example.com" alt="Mrs Hudson" width="300" />}
+      assert markdown == reformat(markdown)
+    end
+
+    test "oneline comment" do
+      markdown = "<!-- The Game, Mrs Hudson, is on! -->"
+      assert markdown == reformat(markdown)
+    end
+
+    test "multiline comment" do
+      markdown = """
+      <!--
+      The Game, Mrs Hudson,
+      is on!
+      -->\
+      """
+
+      assert markdown == reformat(markdown)
+    end
+
+    test "ruler" do
+      markdown = "---"
+      assert markdown == reformat(markdown)
+    end
+
+    test "paragraph" do
+      markdown = """
+      First paragrpah.
+
+      Second paragraph.\
+      """
+
+      assert markdown == reformat(markdown)
+    end
+
+    test "heading" do
+      markdown = """
+      # Heading 1
+
+      ## Heading 2
+
+      ### Heading 3
+
+      #### Heading 4
+
+      ##### Heading 5
+
+      ###### Heading 6\
+      """
+
+      assert markdown == reformat(markdown)
+    end
+
+    test "code block" do
+      markdown = """
+      ```elixir
+      Enum.to_list(1..10)
+      ```\
+      """
+
+      assert markdown == reformat(markdown)
+    end
+
+    test "blockquote" do
+      markdown = """
+      > The Game, Mrs Hudson,
+      > is on!\
+      """
+
+      assert markdown == reformat(markdown)
+    end
+
+    test "table with header" do
+      markdown = """
+      | State | Abbrev | Capital |
+      | ----: | :----: | ------- |
+      | Texas | TX     | Austin  |
+      | Maine | ME     | Augusta |\
+      """
+
+      assert markdown == reformat(markdown)
+    end
+
+    test "table without header" do
+      markdown = """
+      | Texas | TX | Austin  |
+      | Maine | ME | Augusta |\
+      """
+
+      assert markdown == reformat(markdown)
+    end
+
+    test "basic unordered list" do
+      markdown = """
+      * Olafur Arnalds
+      * Hans Zimmer
+      * Philip Glass\
+      """
+
+      assert markdown == reformat(markdown)
+    end
+
+    test "spaced unordered list" do
+      markdown = """
+      * Olafur Arnalds
+
+      * Hans Zimmer
+
+      * Philip Glass\
+      """
+
+      assert markdown == reformat(markdown)
+    end
+
+    test "unordered list with block items" do
+      markdown = """
+      * Quote
+
+        > We can't control what happens to us, only how it affects us and the choices we make.
+
+      * Code
+
+        ```elixir
+        Enum.to_list(1..10)
+        ```\
+      """
+
+      assert markdown == reformat(markdown)
+    end
+
+    test "nested unordered list" do
+      markdown = """
+      * Olafur Arnalds
+        * Particles
+        * Doria
+      * Hans Zimmer
+        * Time
+      * Philip Glass
+        * Opening
+        * The Poet Acts\
+      """
+
+      assert markdown == reformat(markdown)
+    end
+
+    test "basic ordered list" do
+      markdown = """
+      1. Olafur Arnalds
+      2. Hans Zimmer
+      3. Philip Glass\
+      """
+
+      assert markdown == reformat(markdown)
+    end
+
+    test "spaced ordered list" do
+      markdown = """
+      1. Olafur Arnalds
+
+      2. Hans Zimmer
+
+      3. Philip Glass\
+      """
+
+      assert markdown == reformat(markdown)
+    end
+
+    test "with block items ordered list" do
+      markdown = """
+      1. Quote
+
+         > We can't control what happens to us, only how it affects us and the choices we make.
+
+      2. Code
+
+         ```elixir
+         Enum.to_list(1..10)
+         ```\
+      """
+
+      assert markdown == reformat(markdown)
+    end
+
+    test "ordered list: nested" do
+      markdown = """
+      1. Olafur Arnalds
+         1. Particles
+         2. Doria
+      2. Hans Zimmer
+         1. Time
+      3. Philip Glass
+         1. Opening
+         2. The Poet Acts\
+      """
+
+      assert markdown == reformat(markdown)
+    end
+
+    test "raw html" do
+      markdown = """
+      <div class="box" aria-label="box">
+        Some content
+      </div>\
+      """
+
+      assert markdown == reformat(markdown)
+    end
+
+    test "separates blocks with a single line" do
+      markdown = """
+      The first paragraph,
+      with multiple lines.
+
+      > We can't control what happens to us,
+      > only how it affects us and the choices we make.
+
+      ```elixir
+      Enum.to_list(1..10)
+      ```
+
+      Another paragraph.\
+      """
+
+      assert markdown == reformat(markdown)
+    end
+
+    # By reformatting we can assert correct rendering
+    # by comparing against the original content.
+    defp reformat(markdown) do
+      {:ok, ast, []} = EarmarkParser.as_ast(markdown)
+      MarkdownHelpers.markdown_from_ast(ast)
+    end
+  end
+end