Free PDF to Excel Table Extractor

Yes. When a page has no readable text layer, it goes to an AI vision model that reads the image and rebuilds the rows and columns. This page converts up to 3 scanned pages per upload.

For longer scanned documents, PDF2Excel has no page limit and does the same work offline on your own PC.

On a typed PDF, very accurate: the real text coordinates are there to read, so nothing is being guessed. A PDF written by Word, Excel or accounting software falls into this group.

On a scan it depends on the scan. Clean, straight, high-contrast pages come out well. Merged cells, nested tables and unusual layouts are the hard cases anywhere, and this is where you will see the difference.

No. Your PDF is held in memory while the tables are pulled out, and deleted the moment that is done. Nothing is kept on disk and nothing is passed to anyone else.

If a document is confidential enough that uploading it at all is the problem, use PDF2Excel instead. It runs on your own PC and never touches the internet.

PDFs up to 5 MB. Typed PDFs up to 30 pages, scanned PDFs up to 3 pages per upload, since reading a scan takes real computing time. There is also a limit of 3 conversions a day per visitor, which resets at midnight UTC. The counter under the drop zone shows where you are.

PDF2Excel has none of these limits, because the work happens on your own machine.

No, and neither can the desktop app. Encryption hides the page content before any extractor gets to see it. Unlock the PDF first with whatever you normally use, then convert the unlocked copy here.

Usually one of three things: the scan quality is too low for the AI to separate the columns, the table has heavily merged cells or rotated text, or several tables overlap on one page and the detector cannot tell them apart.

Two things to try before giving up. Turn on Always read as images in the advanced options, which sometimes finds tables the direct method misses. Or lower the detection threshold, which makes it look harder for tables without borders. PDF2Excel carries extra table engines and does better on the awkward ones.

Not here. One PDF per conversion, 3 a day. If you do this regularly, PDF2Excel takes a whole folder and converts it in one pass, offline.

Yes. Statements and invoices that come as typed PDFs convert well here, and joining tables across pages is on by default, which is what a multi-page statement needs.

For scanned receipts and long statements where every line matters, PDF2Excel is the better fit: no page cap on the AI reading, and nothing leaves your PC.

English, Chinese, Japanese, and most Latin-script European languages: German, French, Spanish, Italian, Portuguese, Dutch and Polish. Accented characters and scripts beyond those are less reliable. There is nothing to pick, the model reads whatever it sees.

Typed PDFs are a different matter entirely: the text is taken straight from the file, so any language works.

Yes, by default. Every table is written as a native Excel table with filter buttons, a styled header and banded rows, ready the moment you open the file. Turn off Make real Excel tables in the options if you would rather have plain cells.

It is a fair option. Power Query's PDF connector works on Windows Excel 2019 and later, and if your PDF has a text layer and you are already in Excel, it does the job.

Use this page instead when the PDF is a scan, because Power Query has no way to read one; when you are on a Mac, on Linux, or on an older Excel; or when you would rather drag, drop and download than click through a menu and preview 29 query results one at a time.

The output here is a native Excel table too, so you end up with the same thing to work with either way.