PDF to Markdown Converter

Markdown is the plain-text format used by GitHub, Notion, Obsidian and many note apps. Turning a PDF into Markdown makes it easy to edit, version and paste into a knowledge base. This converter reads the text layer of the PDF in your browser and rebuilds headings, bold and italic, lists, links and paragraphs, joining wrapped lines and hyphenated words again.

How to use this tool

1. Drop one or more PDF files (scanned PDFs need OCR first).
2. Optionally limit the pages and choose what goes between pages.
3. Keep "Drop repeated headers and footers" on to remove page numbers and running titles.
4. Copy the Markdown from the text box or download the .md file.

Formula

Body size = the font size with the most characters. Sizes more than 15% larger become # , ## , ### and #### from the largest down.

Lines are joined into one paragraph when they share a size, their baselines are less than 1.9 × the font size apart and they start at about the same x position, unless the previous line ended early with a full stop. A line ending in a hyphen followed by a lowercase word is joined without the hyphen. Lines starting with •, –, * or 1. become list items. Text near the top or bottom 8% of the page that repeats on at least half the pages is treated as a header or footer.

Example

A report page with "Annual Report 2026" in 24 pt, "Overview" in 16 pt and body text in 11 pt becomes # Annual Report 2026, ## Overview and a paragraph. A sentence wrapped as "con-" / "tinues" becomes "continues", bold words become bold words, "• First item" becomes - First item, and a four-column price list becomes a table starting | SKU | Item | Price | Stock |.

Tables and limits

Consecutive lines whose words are separated by wide gaps are written as a Markdown table, with the first row as the header; merged cells or cells that wrap onto two lines may need a quick fix by hand. A PDF stores positioned characters, not structure, so the result is a careful reconstruction: images are left out, pages set in two columns are read column by column, and more complex layouts (three columns, sidebars) may interleave. Fonts that do not name their weight are treated as regular text.