Extract tables from a PDF to Excel

Open the PDF in PDFCells. It finds the tables on every page, joins tables that run across pages, checks totals and cross-totals, and lets you fix any cell before downloading an .xlsx with real numbers and dates.

Updated October 21, 2026

Steps

1. Go to pdfcells.com and choose the PDF, or drop it on the page. Drop several files to open them as a batch. A password-protected PDF asks for its password.
2. Every page is pre-checked and gets a read method: PDF text for pages with a text layer, OCR for scans, or AI for pages you choose.
3. Review each table next to its page. Mark an area to keep or ignore when a page has more than you need, and set header rows if they were missed.
4. Choose a sheet layout: one sheet per table, one per page, or all tables on one sheet. Download.

What “accurate” means here

• Characters: pages with a text layer are read from the PDF’s own text, so there are no OCR typos. Scans use OCR, which can misread.
• Structure: rows and columns are rebuilt from where words sit on the page. Side-by-side tables and multi-level headers are the hardest layouts; check those in the grid.
• Arithmetic: totals must foot, columns such as Q1–Q4 must cross-foot to the year, quantity × price must equal the line total, and running balances must tie. A table whose totals all add up is marked Footed.

In the Excel file

• Numbers, percentages, currencies, negatives in parentheses and full dates become real Excel values with consistent column formats.
• Header rows are frozen. Tables continuing onto the next page are joined into one table.
• Interior subtotals sit below the Excel Table, so pivot tables don’t double-count them.
• An optional Audit sheet records how each page was read, which checks passed and which cells you edited.

Limits and cost

One file can have up to 300 pages. A batch holds up to 200 files and 2,000 pages in all.

Reading the PDF text and OCR are free, need no account and run in your browser. AI reading is optional: signed-in users get 25 free AI pages a day on the Fast model, then prepaid credits at Anthropic's token cost plus 15% (packs of $10, $25, $50), or no markup with your own Anthropic API key.

Questions

Can it extract a table that spans several pages?

Yes. Tables that continue onto the next page with the same columns are joined into one table and one sheet.

Does Excel’s own Get Data › From PDF do the same?

Excel’s PDF connector imports tables from PDFs with a text layer into Power Query, and is a good choice if you already use Power Query. PDFCells adds OCR for scans, the checks and an editable grid next to the page.

What PDFs don’t work well?

Pages with side-by-side tables, multi-level headers or very poor scans need the most checking. Use areas to separate tables, or AI for that page.