Extract Table From PDF

Looking for a way to extract table from PDF pages without retyping them, even when the PDF is a scan? Upload it and Linnk reads every page, finds each table and writes it into an Excel workbook: one sheet per table, named after the page it came from, with merged header cells kept. A CSV per table, a layout spreadsheet and a searchable PDF come from the same run, and the text stays in the language it was written in.

Output format

ExcelMarkdownSearchable PDFPlain text

You'll also get CSV · Spreadsheet (layout) · Searchable PDF

Output language

Drop a document to convert

Accepts scans and photos — PDF, JPG, PNG, HEIC

Free preview · No sign-up to try

Encrypted & private 154 languages & dialects Tables kept as tables Files deleted after 72h

How do you extract a table from a PDF, even a scanned one?

Upload the file to Linnk's Extract Table From PDF tool and it writes every table it finds into an Excel workbook, whether the PDF was exported from software or scanned on a copier. The second case is the hard one. Many PDF table tools read the file's hidden text layer, and a scanned page has none; it's only a picture. Linnk reads the page image itself: AI models (Linnk uses ChatGPT, Claude and Gemini models) rebuild each page as structured text, pick out the tables and write them into the workbook with one sheet per table, sheets named by page and merged header cells kept as real merges. The same run gives you one CSV per table, a layout spreadsheet that looks like the page and a searchable PDF. Amounts, dates and account numbers are copied digit for digit, and a digit the models can't make out is marked with a question mark instead of being guessed. Two limits are worth knowing: a table that runs across several pages comes out as one sheet per page rather than one joined sheet, and a PDF with no table gets a note saying so, not invented rows. A straight, well-lit scan usually reads best; faint or skewed pages may need a quick check.

1 sheet per tableExcel workbook, named by page

Scanned or digitalPDFs read from the page image

CSV includedone file per table, UTF-8

Freepreview, no account needed

Last updated: October 2026

How to extract table from PDF files into Excel

Table extraction from PDF online in three steps, with nothing to install.

Upload the PDF

Drop in a scanned or digital PDF. Excel is already selected as the output, and the output language is set to Keep original. Every page is read as an image, so a copier scan goes through the same process as a report exported from software.

Check the first table

Usually within a minute per page, the preview shows the first table as a grid, up to a row limit, and copying the recognised text is free. The other tables are in the download. Look at any cell with a question mark first; that's a digit the scan didn't show clearly.

Download the Excel workbook

Sign in with a free account to download. The CSV files (one per table, zipped when there are several), the layout spreadsheet and the searchable PDF come from the same conversion, and so do Markdown and plain text. If you need a Word file, it's built on request in a few extra minutes.

Built to extract table from PDF scans, not only clean exports

Bank statements, invoices, annual reports, lab reports and archive pages that started life on paper.

One sheet per table, named by page

An annual report with a balance sheet on page 12 and a cash-flow table on page 14 gives you two sheets, each named after its page, so you can go straight back to the source when a figure looks off. Merged header cells stay as real merges, so a two-level header looks the way it does in the PDF.

Scanned PDF to Excel, not only digital

Many table tools depend on a text layer that a scan doesn't have. Linnk reads the page image, so faxed statements, copier scans and old archive pages go from a paper scan to Excel the same way an exported report does. A straight, clean scan reads best.

Digits copied, not guessed

Amounts, account numbers, dates and quantities are copied digit for digit. A digit too faint to read is marked with a question mark and an unreadable passage is marked [illegible], so you know which cells to check before you total a column.

No invented tables

If a PDF holds no table, the Excel workbook and the CSV say so instead of making rows up. You still get the layout spreadsheet, the searchable PDF, Markdown and plain text from the same upload.

Keep the language, or translate it

Keep the original language or translate in the same run. By default a German supplier statement stays in German, and Arabic, Hebrew, Chinese, Japanese and Korean tables are read in their own script.

CSV and a layout copy from the same upload

Next to the Excel file you get one CSV per table in UTF-8, so accented and non-Latin text opens correctly, plus a layout spreadsheet with one sheet per page that looks like the original and a searchable PDF rebuilt as real text.

PDF table extraction facts

What to expect before you upload a PDF.

Input

PDFs: Scanned or digital
Images: JPG, PNG, WEBP, HEIC, HEIF
Not accepted: Word, PowerPoint, Excel
Language: Kept as is by default

Output

Excel (tables): One sheet per table, named by page
CSV: One file per table, UTF-8
Spreadsheet (layout): One sheet per page, laid out like it
Table across pages: One sheet per page, not joined

Accuracy

Numbers and dates: Copied digit for digit
Unreadable digits: Marked with ?, not guessed
Merged cells: Kept as real merges
PDF with no table: Reported, not invented

Speed and access

Speed: Usually under a minute per page
Preview: Free, no account
Download: Free account
Longer documents: Paid plans, see linnk.ai/pricing

Ways to extract tables from a PDF, compared

How Linnk lines up against the usual ways of getting a PDF table to Excel.

Feature	Linnk Extract Table From PDF	Copy and paste	Text-layer table tools	Spreadsheet app PDF import
Scanned PDFs	Read from the page image	Usually nothing to select	Usually need OCR first	Usually need a text-based PDF
Several tables in one PDF	One sheet per table, named by page	One at a time, by hand	Usually picked one by one	Picked from a list of detected tables
Merged header cells	Kept as real merges	Usually lost	Varies	Varies
Unclear or unreadable digits	Marked with a question mark	Your own judgement	Copied from the text layer on digital PDFs	Copied from the text layer on digital PDFs
Table running across pages	One sheet per page, not joined	Joined by hand	Varies	Varies
Language of the table	Kept, or translated in the same run	Kept	Kept	Kept
Other outputs from the same file	CSV, layout spreadsheet, searchable PDF, Markdown, text	None	Varies, often CSV	The workbook only

Comparison as of September 2026. Other tools vary by version and settings; check each one for its current options.

Who extracts tables from PDFs with Linnk

People with numbers locked in a PDF and a spreadsheet waiting for them.

Accountants and bookkeepers

Pull transaction tables out of scanned bank and card statements for reconciliation. Each statement page lands on its own sheet, ready to paste into one ledger.

Financial analysts

Move the balance sheet and income statement from an annual report PDF into a model instead of retyping the figures, then check the flagged cells.

Accounts payable and procurement

Turn supplier invoices and price schedules that arrive as scanned PDFs into rows you can match against purchase orders.

Researchers and students

Get data tables out of older journal articles and theses that exist only as scans, so the numbers can go into a spreadsheet or a statistics package.

Auditors and compliance teams

Copy tables from scanned filings, permits and inspection registers into a workbook for sampling, with the source page in each sheet name.

Importers and logistics teams

Convert packing lists, rate sheets and customs documents sent as PDFs into Excel in their original language, so product names match what the supplier wrote.

Extract Table From PDF FAQ

How do I extract table from PDF pages with Linnk?

Upload the PDF, scanned or digital. Excel is preselected, so Linnk reads each page, shows the first table as a grid in a free preview and lets you download the full workbook once you sign in with a free account. Each table lands on its own sheet, named after the page it came from.

Does it do PDF to Excel OCR on scanned files?

Yes. Instead of depending on the PDF's text layer, AI models read each page image, so you can convert a scanned PDF to Excel the same way as a digital one. If the tables are still on paper, scan them to PDF first and upload that. A straight, well-lit scan usually reads best; faint, curled or skewed pages may need a quick review.

Why use Linnk instead of copying the table by hand?

Copying a table out of a PDF viewer often runs the columns together, and on a scan there's usually nothing to select at all. Retyping a three-page statement can take an hour, and one slipped digit is easy to miss. Linnk usually reads each page in about half a minute, keeps merged headers and the table's own language, and marks any digit it can't read, so you're checking rather than typing.

What happens to a table that runs over several pages?

It isn't joined. Each page's part of the table comes out as its own sheet, and its own CSV, named by page. A long statement therefore gives you one sheet per page; paste them under each other in Excel to rebuild the full table, and delete any header row that repeats.

What if my PDF has no tables?

Tables are never invented. If the PDF holds no table, the Excel and CSV rows on the result page say so, and you still get the layout spreadsheet, the searchable PDF, Markdown and plain text from the same upload.

How accurate are the numbers?

Amounts, IDs, dates and account numbers are copied digit for digit from the page. A digit that can't be read is marked with a question mark and an unreadable passage is marked [illegible], rather than guessed. Results still depend on the scan, so look over the flagged cells and check a column total against the PDF before you rely on it.

Which files can I upload?

PDFs, scanned or digital, are what this page is built for. JPG, PNG, WEBP, HEIC and HEIF images are accepted too and read as one-page scans; for phone photos and screenshots, the Image to Excel tool is set up for that job. Word, PowerPoint and Excel files aren't accepted, because they already contain their text.

Do I get a CSV file as well?

Yes. The same run writes one CSV per table, zipped when there are several, in UTF-8 so accented and non-Latin characters open correctly. You also get a layout spreadsheet with one sheet per page that looks like the original, which helps with forms where the table is only part of the page.

Will the tables be translated?

Not unless you ask. The output language is set to Keep original, so a French invoice comes out in French. To get the tables in another language, switch the setting to Translate to… before you start, and the extraction and the translation happen together. Convert reads the most widely used languages and scripts, including Arabic, Hebrew, Chinese, Japanese and Korean.

How long does it take?

Usually under a minute per page, and roughly half a minute on a typical scan. The page shows an estimate in minutes while it works. A Word file, if you ask for one, takes a few extra minutes because it's built after the conversion.

Is it free to extract tables from a PDF?

The preview is free without an account, the first pages of a document are converted for free, and copying the recognised text is always free. Downloading the Excel file and the other formats needs a free account. Longer documents are covered by the paid plans; see linnk.ai/pricing for the current options.

What happens to my PDF after the conversion?

Converted files are deleted automatically after 72 hours, and the result page tells you so. If a statement or report is confidential, download what you need within that window and check your organisation's rules before uploading.

Extract table from PDF files, scanned or digital

Upload a PDF and look at the first table as a grid before you download anything.

Extract tables to Excel View pricing