How to Extract Tables from a PDF to CSV Online for Free

Direct answer: You can extract tables from a PDF to CSV online for free with the Extract Tables from PDF tool: it reconstructs tables geometrically from positioned text, previews every table on screen with its page number and size, and lets you download the result as CSV files that open in Excel, LibreOffice, or Google Sheets. A 12-page price list, for example, can become a sortable spreadsheet in about a minute.

The value is in what a spreadsheet lets you do next. A table trapped in a PDF can be read, but it cannot be sorted, filtered, totalled, or charted without retyping it. Once the same table is a CSV, you can sort a price list by price, filter a statement by category, sum a column, or paste the data into a report, all without retyping a single row. That is why table extraction is often the first step in turning a published PDF into working data.

How tables are reconstructed from positioned text

A PDF does not store a table as a table. It stores individual words at precise positions on a page, and a table is only implied by how those words line up. The tool reconstructs that implied structure geometrically: words that share a baseline belong to the same row, and vertical gaps between word groups mark column boundaries. By applying those two rules across a page, it rebuilds rows and columns that match the visual table a human reader sees.

Understanding that method explains both the strengths and the limits of extraction. A clean, aligned table with consistent columns and clear spacing reconstructs well, because its geometry is unambiguous. A table with merged cells, irregular spacing, or no visible borders is harder, because the geometry alone does not always say where one cell ends and the next begins. The tool does its best with such layouts, but the result may need tidying, which is why previewing before download is an essential part of the workflow, not an optional extra.

Tables only or everything: choosing the output

The tool offers two output modes, and choosing between them depends on what you actually need. Tables only extracts just the tables, skipping prose pages, and gives you either one combined CSV containing all tables, or one CSV per table in a ZIP. That mode suits a price list, a catalogue, or a report where the tables are the data and the surrounding text is not needed in the spreadsheet. One combined CSV is convenient when the tables share a structure, while one CSV per table keeps differently structured tables from being forced into a single file.

Everything mode treats every page as a grid and exports the whole document's positioned text in grid form. That mode suits a document where the prose itself is arranged in a grid-like way, or where you want a complete page-by-page record rather than only the formal tables. It will produce more output, including pages you might not need, so use it deliberately when completeness matters more than tidiness. For most everyday tasks, start with Tables only, and switch to Everything only if the tables you need are not being captured on their own.

Step by step: extracting your tables

Load the PDF and let the tool detect its tables, then work through the preview before downloading anything. Each preview shows the table, its page number, and its size, so you can tell at a glance whether a table was captured completely, whether a column was split or merged incorrectly, and whether a table you expected is missing. That preview step is where most errors are caught, and it takes far less time than fixing a downloaded CSV after the fact.

When the previews look right, choose your output mode and download. Open the CSV in your spreadsheet program and check the same things again at spreadsheet level: column headings, row counts, and any cell that looks merged, split, or shifted. A quick sort or filter test will reveal a misaligned column immediately, because sorted values will appear in the wrong place. Fixing issues at this stage usually means adjusting your expectations or tidying a few cells, not starting over.

• Open the free tool and load the PDF containing your tables.
• Review every table preview, checking its page number and size.
• Look for split columns, merged cells, or missing rows in the preview.
• Choose Tables only or Everything, as your task requires.
• Download one combined CSV, or one CSV per table in a ZIP.
• Open the CSV in Excel, LibreOffice, or Google Sheets and verify it.
• Sort, filter, or total the data, and tidy any cells that need it.

When extraction needs tidying, and when scans need OCR first

Clean aligned tables extract well, and that covers a large share of real documents: price lists, timetables, financial statements, and product catalogues that were produced from a spreadsheet or a typesetting system. Merged cells, borderless layouts, tables with multi-line cells, and tables where columns are separated only by wide spacing may need tidying after download. Tidying usually means splitting a merged cell, moving a value into the correct column, or deleting a stray header row that was repeated across pages. Budget a few minutes for that work on complex documents, and treat a perfectly clean first download as a pleasant surprise rather than a guarantee.

Scanned pages are a different case entirely, and they need a different first step. A scan has no text layer, so there are no positioned words for the geometric method to work from, and table extraction will not work on a scan as it stands. Run OCR first to build a text layer over the scanned pages, and only then extract tables from the searchable result. That two-step workflow, OCR then extraction, is the correct route for scanned price lists, scanned statements, and scanned reports, and it will save you from puzzling over an empty extraction from a file that simply had no text to extract yet.

To understand the OCR step in full, you can read our guide to making scanned PDFs searchable, and for the reverse workflow, turning images into a PDF in the first place, see more guides on further Asad Webs guides. Keeping the workflow order straight, scan to OCR to extraction to spreadsheet, is what makes each step work as intended.

Working with the CSV in a spreadsheet

CSV is a plain, universal format, which is why it is the right output for this task. It opens in Excel, LibreOffice, and Google Sheets without special software, and it can be imported into databases, accounting tools, and data tools just as easily. When you open a CSV, check that your spreadsheet program has interpreted the columns correctly, especially where values contain commas, currency symbols, or dates, because a misinterpreted column can look right at first glance while sorting incorrectly underneath.

Once the data is in a spreadsheet, the usual spreadsheet skills apply: sort by any column, filter to the rows you need, add totals, and format headings for readability. Save the working copy in your spreadsheet program's native format after tidying, and keep the original CSV as a clean export from the PDF, so that you can always return to the unedited extraction if a later edit goes wrong. That habit, original export plus working copy, protects you against both accidental overwrites and the slow drift that happens when a tidied file is edited many times. When a value contains a comma inside a quoted field, a careful spreadsheet import will keep that value in one cell, whereas a naive split on every comma would break it into two cells and shift every later column on that row, so checking a few comma-containing rows after import is a worthwhile habit. If a column of dates sorts alphabetically instead of chronologically, check the date format in the CSV import step, because a spreadsheet that treats dates as text will order them by characters rather than by calendar order, and converting the column to a proper date format before sorting will restore the correct sequence.

Frequently Asked Questions