Turn PDF tables into a spreadsheet
Drop a PDF here or choose a file. It is read on your own device, so nothing is uploaded and there is no artificial page limit - practical limits depend on your browser and device memory. Export to Excel, CSV, TSV, JSON, Markdown or HTML.
Detection
Lower sensitivity means fewer, wider columns. Drag a guide on the page to move a column, double click a gap to add one, and double click a guide to remove it.
Clean up
Pages and output
Text file options
Title
PDF to Excel and CSV converter
Pull the tables out of a PDF and get a spreadsheet you can actually work with. Your file is opened by your own browser, so it is never uploaded, there is no artificial page limit, no watermark and no sign-up. Drop a PDF in, check the columns the converter found, fix anything it got wrong, and download Excel, CSV, TSV, JSON, Markdown or HTML.
Why PDF tables are hard, and what this does about it
A PDF does not store a table. It stores characters at coordinates, and the rows and columns you see are an illusion created by where those characters sit. Every converter has to reconstruct the structure, and that is where most of them quietly fail: they hand you one merged column, or split a name into three, and offer no way to fix it.
This converter gives you the controls instead of hiding them. It runs two detection engines and shows you which one it used:
- Ruled lines reads the actual border lines drawn in the PDF and uses them as the grid. It is exact on invoices, statements and equipment registers that have visible borders, and it handles cells whose text wraps onto two or three lines without splitting the row.
- Whitespace is for tables with no borders. It measures the vertical gaps that run down the whole block and treats a gap that never gets filled as a column boundary.
- Text lines is the fallback that gives you one line of the page per row, useful when the layout is not really a table at all.
If a guess is wrong, drag a column guide on the page preview, double click a gap to add a guide, or double click a guide to remove it. You can also select just the area of the page that holds the table and ignore everything else. The data grid updates as you go, and every cell is editable before you download.
Built for real documents
Tables that run across pages
Choose to combine pages into a single sheet and the pages are joined in order, with the header row that repeats at the top of each page removed automatically.
Numbers that behave like numbers
Currency symbols, thousands separators, percentages and accounting negatives written as (123.45) are recognised and written to Excel as real numeric cells, so your totals and formulas work the moment the file opens. Codes such as V-101, 6-P-1102 and account numbers are deliberately left as text so they are not mangled.
Excel files that are ready to use
The workbook is written with a styled header row, the header frozen, autofilter switched on and column widths sized to the content. Each table can go to its own sheet, one sheet per page, or all into one.
Rotated, landscape and multi-table pages
Pages rotated by ninety degrees are straightened before detection, and a page holding two or three separate tables produces separate tables rather than one scrambled block.
Scanned pages
A scan is a picture, so there is no text to extract. When a page has no text layer the converter says so and offers optional English OCR, which reads the words from the image in your browser and feeds them through the same detection. The bundled training data covers English only, so a scan in another language will come back badly. OCR is slower and less reliable than a real text layer, so always check the result against the page.
Password protected files
If the PDF needs a password to open, you are asked for it here and it is used locally to unlock the file. The password is not stored or sent anywhere.
Private by design
There is no upload step, because there is nothing on the other end to upload to. Your PDF is parsed by your browser and the spreadsheet is generated on your device. This is not the same as the tool being fully offline, and it is worth being precise about the difference. Your browser does reach out over the network to fetch program code: the PDF reading library, the font data files it needs for documents that do not embed their own fonts, and, only if you ask for it, the OCR engine and its English training data. Those requests carry code and typefaces towards you. Nothing about your document travels the other way. No page image, no extracted text, no OCR output and no generated spreadsheet ever leaves the machine. If you want the network out of the picture entirely, self-host the reading library and point the tool at your own copy.
That is also why there are no artificial limits. Services that process your file on their servers have to cap pages, file size or daily conversions to control cost, and many keep your document for a period afterwards. Here the work happens on hardware you already own, so nothing is metered or capped on our side. Be clear about what that does and does not promise: it removes the policy limit, not the physical one. Practical limits depend on your browser and device memory, and a long document full of dense tables will still be slow on a phone or an older laptop. For those, set a page range and convert in blocks rather than taking the whole file in one pass.
A safety detail most converters miss
A spreadsheet treats a cell beginning with an equals sign, a plus, an at sign or a minus as a formula, and opens it as executable content. A PDF you were sent by someone else can carry exactly that in a cell. This converter neutralises those cells by default so the exported file cannot run anything when it is opened, while still keeping genuine negative numbers intact. You can switch it off if you know the source is trustworthy.
How to convert a PDF table to Excel
1. Open your PDF
Drop it on the page or choose it from your device. Nothing is uploaded.
2. Check what was found
The detected table is outlined on the page preview with a guide on each column boundary. Switch the engine or drag the guides if the split is not right.
3. Tidy the data
Edit any cell, delete a stray row or column, mark the header, and turn on joining for rows that wrapped onto a second line.
4. Download
Take the Excel workbook, or a CSV with the separator and encoding you need. You can also copy the whole table to the clipboard and paste it straight into a sheet.
Who it is for
Accountants and bookkeepers lifting line items off supplier invoices and bank statements. Analysts pulling figures out of annual reports and regulatory filings. Engineers converting equipment registers, valve schedules and inspection reports into a working list. Researchers extracting results tables from published papers. Anyone who has been handed a PDF when what they needed was a spreadsheet.
Frequently asked questions
Is my PDF uploaded to a server?
The PDF itself is never uploaded: it is opened, parsed and converted by your own browser, and the spreadsheet is written on your device. Your browser does fetch program code over the network - the PDF reading library and the font data files it needs for documents that do not embed their own fonts, and, only if you ask for OCR, the OCR engine and its English training data. Those downloads are code and typeface data going one way, to you. No page image, no extracted text and no file bytes ever go the other way.
Is there a page limit or a file size limit?
There is no artificial page limit and no daily conversion limit, because there is no server doing the work and nothing to meter. That is not the same as unlimited capacity: the practical limit is the memory and speed of your own browser and device, and a few hundred pages of dense tables will tell on a modest laptop or a phone. Nothing is capped on our side, but if a large document feels slow or the tab runs out of memory, convert it in blocks by setting a page range such as 1-25 and working through it.
Why do some tables come out wrong?
A PDF stores characters at coordinates, not rows and columns, so every converter has to infer the structure. If the guess is wrong, switch the engine between ruled lines and whitespace, move the column guides by hand, or select just the table area on the page. The preview updates as you change things.
Can I convert several PDFs at once?
Yes. Choose or drop as many as you like and each one is opened and converted in the same pass. The Excel workbook contains the tables from every loaded PDF, with the source file name in each sheet name, so a folder of invoices becomes one workbook. The picker above the page preview switches between the loaded documents, and switching does not discard anything already extracted. Everything still happens in your browser, so the practical ceiling is the memory of your device rather than a policy limit.
How does it decide whether 1,234 is a thousand?
It uses the pattern of the separators rather than assuming the last one is the decimal point, and Auto looks at the extracted tables to determine whether numbers use 1,234.56 or 1.234,56. That is what decides the genuinely ambiguous cases: 1.234 is one thousand two hundred and thirty four where the tables also contain 1.234,56, and one point two three four where they contain 1,234.56. With several PDFs loaded the evidence from all of their tables is read together. Set the format explicitly if there is too little evidence or if the files do not share one convention. Identifiers such as V-101, 6-P-1102 or 10-20 are left as text so a tag number is never turned into a date or a number.
Does it work on scanned PDFs?
A scanned page is a picture with no text in it, so there is nothing to extract directly. When the converter finds a page with no text layer it offers optional English OCR, which reads the words from the image in your browser and then runs them through the same table detection. Only English is included. OCR is slower and less exact than a real text layer, so always check the result.
Can it handle a table that runs across several pages?
Yes. Choose to combine pages into one sheet and the converter joins the pages in order and removes the header row that repeats at the top of each page.
Are numbers exported as real numbers?
Yes. Currency symbols, thousands separators, percentages and accounting negatives in brackets are recognised and written to Excel as numeric cells, so totals and formulas work straight away. You can turn this off if you want every cell kept as text.
Can I edit the data before downloading?
Yes. Every cell in the preview is editable, and you can delete rows and columns, transpose the table and undo your changes. What you see is exactly what gets exported.
Can I embed it in WordPress?
Yes. It is a single HTML file with all styles scoped to one container, so it drops into a WordPress shortcode, a Gutenberg or Elementor HTML block, or a standalone host such as Cloudflare Pages, Vercel or Netlify.