Add page numbers to a PDF

How it works

The number is written straight into each page's content stream, the same list of drawing instructions that holds the rest of the page. That has one consequence worth understanding before you run it: the number becomes part of the document. It is not a header or footer field of the kind a word processor maintains, so there is nothing for a later reader to switch off and nothing for a later tool to find and strip. Every viewer shows it, every printer prints it, and text extraction reads it as ordinary text alongside the body copy.

What survives the edit

Nothing else about the file is disturbed, because the document is edited in place rather than rebuilt out of its pages: fillable fields keep their names and the values already typed into them, checkboxes keep their state, and bookmarks still jump where they used to.

A form or statement that is encrypted only against printing or editing is read and numbered normally, which covers most of what arrives from a tax authority or a bank; a PDF that needs a password just to open is refused with an explanation instead, because this tool never asks you for a password.

And if you might need an unnumbered copy, keep the original file. That is the only reliable route back, and it costs nothing, because the original is still on your disk exactly as it was.

Printed number versus physical sheet

The starting number and the skip-first-page switch exist because the printed number is often not the physical sheet. A report with a cover and a contents page has two sheets that no reader thinks of as page one, so leaving the cover unnumbered and starting the count after it is the ordinary convention rather than an edge case.

Starting somewhere other than 1 covers the other common situation: a file that continues a document which ended at page 11 should begin at 12, so the reader's numbering runs straight through the join. When the wording includes a total, that total is the last number actually printed rather than the sheet count, which is why the final page always agrees with its own total instead of reading "23 of 24".

Helvetica and 244 bytes a page

Text is drawn in Helvetica through the PDF standard-font mechanism. The fourteen standard fonts are named in the specification and every conforming viewer supplies them, so no font file is added: the document gains one font reference shared by every page, plus one short drawing instruction per page. Measured across two real 130-page books, that instruction costs 244 bytes per page.

Why the saved file grows

The size change reported to you afterwards is a different number, sometimes a much larger one, and it is worth expecting, because saving a PDF rewrites the whole file rather than appending to it. Where the original stored its cross-reference table and its objects as compressed streams, the rewrite writes them back out plainly: a 441 KB, 127-page book measured here came back as 854 KB, of which the page numbers were only 30 KB.

The rewrite runs in both directions, and that same structure predicts which one. Across the real documents tested here, files not using it came back anywhere from 36% smaller to 1% larger, because a rewrite also drops whatever slack the original left behind. Neither direction loses any detail: gzipped, input and output land within 5.5% of each other in every case, shrinking or growing. The growth is therefore recoverable redundancy, which is why compressing afterwards is a measured recommendation here rather than a generic upsell.

The one tradeoff left is typographic. The number is set in Helvetica rather than your body typeface, which at footer size is rarely noticeable and would otherwise mean embedding a 30 to 100 KB font file.

Alignment measured per page

Alignment is measured rather than estimated. Each page reports its own dimensions, the width of the number is measured in Helvetica at the size you chose, and the coordinates are computed from those two figures. A right-aligned number therefore sits exactly the margin distance from the right edge whether the page is A4, Letter, a landscape spreadsheet or a scanned receipt, and a file that mixes sizes still lines up because nothing is inherited from the first page.

The edge that counts is the one you can see: a print-ready PDF often carries a sheet larger than its visible page, with trim and bleed outside it, and the margin is measured from the visible box rather than the sheet, so the number does not end up out in the trim area.

Pages carrying a rotation flag are placed in the rotated frame you actually see, so the bottom of a sideways scan is the bottom you are looking at. PDF measures from the bottom-left corner rather than the top-left, and the preview above is positioned from the same numbers the transform uses, so where the label sits is where the number lands.