Extract Tables from a PDF into Excel
Every page is examined for rows that line up into the same columns, and each table found becomes its own worksheet named by page and table number. Numbers arrive as numbers so they can be summed straight away, while identifiers that merely look numeric keep their leading zeros as text. If the PDF holds no table at all, the tool falls back to one column of plain text per page rather than failing.
This tool uploads your file — only with your permissionExtract your tables
The tool needs a server to reconstruct a grid, so it asks your permission before uploading and deletes its copy as soon as the workbook is built. Documents up to 25 MB are accepted.
Open PDF to ExcelHow a program knows what a table is
A PDF has no concept of a table. There is no grid object, no row and no cell — only text placed at coordinates and lines drawn where the designer chose. A table exists because those coordinates line up: each column of figures shares a horizontal position, and each row sits on a common baseline. Recognition is therefore an inference from alignment rather than a read of stored structure.
The rule the tool applies is that a run of rows must share the same column positions to be treated as one table, with at least two rows involved. That is exactly how a printed table presents itself, which is why invoices, bank statements, price lists and reports convert without the file declaring anything. It is also why a layout held together only by whitespace is the fragile case: with no ruled lines, two columns that sit close together can be read as one, or a wide gap inside a cell can be read as a column break.
One consequence of reading geometry is that a table splitting in two is not an error. A page carrying two separate tables produces two worksheets rather than one sheet with a gap, because the two grids never shared column positions. The same logic means a totals line at the bottom is ordinary data rather than a header, which is what you want if you intend to recalculate it.
When a PDF contains no table structure at all, the honest output is not an empty workbook. Each page becomes a worksheet with its text down a single column, so you can see immediately that there was no grid to find and decide what to do about it.
Step by step
Add the PDF and approve the upload
Open PDF to Excel and select the document. Rebuilding a grid needs more than the browser can do, so this is one of three tools here that use a server; it explains what happens and waits for your agreement before anything is sent. The limit is 25 MB.
Let the pages be examined
Every page is scanned in order for rows sharing column positions, and a run needs at least two rows before it counts as a table. A long document takes proportionally longer.
Open the workbook and match sheets to pages
Each table becomes its own worksheet, named by page and table number so you can trace it back to the PDF. The first row is treated as a header and shown in bold.
Check the column edges first
Before reading any figure, check that the columns landed where you expect. A table reinforced with ruled lines almost always converts cleanly; one aligned only by spacing is where a split can go wrong.
Confirm the numbers are usable
Values such as 1,240,000 or 12.4% arrive as numeric cells, so you can total them immediately. Identifiers that only look numeric — postcodes, employee numbers, anything with a leading zero — are deliberately kept as text so the zero is not dropped.
Rejoin tables that came out separately
A page with two distinct tables yields two worksheets. That is usually helpful; if you would rather see them together, move or copy the contents within Excel.
Handle the no-table case
If the tool reports no tables, each page has become a worksheet with the text down one column. Extract the figures from there, or run OCR first if the PDF turned out to be a scan with no text layer.
What comes out, by input
The workbook follows the structure the PDF actually has, rather than the structure you might expect it to have.
| The PDF contains | What you get | Worth checking |
|---|---|---|
| Tables drawn with vertical rules | One worksheet per table, cleanly split | Merged cells and multi-line headers. |
| Tables aligned by spacing alone | One worksheet per table, usually | Column edges where two columns sit close together. |
| Two separate tables on one page | Two worksheets for that page | Whether you would rather have them combined in Excel. |
| A table running over several pages | One worksheet per page, so the run is split | Whether the header row repeats on each sheet. |
| No table at all | One worksheet per page, text in a single column | Whether OCR is needed because the file is a scan. |
| A scanned PDF | An error, stated plainly | Run OCR PDF first, then convert. |
Mistakes to avoid
Trusting a table aligned only by whitespace
Ruled lines make edges unambiguous; spacing is a guess. Where two columns sit very close together, or a cell contains a wide gap, the split can land in the wrong place. Compare the first few rows against the original before relying on the figures.
Expecting one worksheet for the whole document
Each detected table gets its own sheet, so a five-page report with a table on every page produces five worksheets. That is by design, and it keeps each grid traceable to the page it came from.
Assuming a scanned PDF will work
A scan is a picture with no text layer, so there is nothing to read and the tool says so rather than returning an empty workbook. Run OCR PDF first, then convert the searchable file.
Overlooking the upload and the size limit
This is a server-side conversion, unlike most tools here, and it accepts documents up to 25 MB. Split a larger file with Split PDF first, then convert the parts.
Reading identifiers as numbers
Postcodes and reference numbers with leading zeros are kept as text precisely so the zero survives. If you need them treated as numbers, convert the column in Excel and accept the lost zero.
Frequently asked questions
Is my PDF uploaded?
How does it decide what counts as a table?
Are the numbers stored as numbers?
How many worksheets will I get?
What if the PDF has no tables?
Why did a column come out merged with its neighbour?
Can it read a scanned PDF?
Is there a file size limit?
What happens to the file after conversion?
Tools used in this guide
Send us feedback
Something broken, or unclear? Tell us and we will look at it.