Add Page Numbers to PDF

It happens every week in every courthouse in the country. Somebody assembles a 90-page exhibit — bank statements, screenshots, a sworn statement — converts it to PDF, files it, and gets it bounced back by the clerk within the hour. The defect is not the content. It is that the pages have no numbers, and without numbers the document cannot be indexed, cited, or referred to in a hearing. 'Reference to Exhibit B, page 41' has to mean something.

Academia runs the same rule with friendlier enforcement. A thesis submitted without continuous page numbering earns a note from the registrar's office and a sheepish resubmission. So does the training manual your boss wants printed and bound, and the board packet, and the grant attachment, and the discovery response. Page numbers look decorative until the exact moment they are structural.

Add Page Numbers to PDF exists for that moment. Drop in the file, choose a position and a format, stamp, download. It runs entirely in your browser — no upload, no account, no watermark on your exhibits — and it handles the numbering conventions that actually get requested: plain Arabic, 'page X of Y', Roman front matter, and Bates-style prefixes for litigation work.

Numbering pages sounds like the simplest possible edit, and mechanically it is. What separates a good stamping tool from a bad one is the options surrounding the mechanical part, because every document culture has its own numbering etiquette.

• Nine positions: left, center, and right across the top, middle, and bottom of the page — the standard grid that legal, academic, and print conventions actually call for.
• Your choice of format: plain numbers, 'Page 3', '3 of 86', Roman numerals, or a Bates-style prefix with zero-padded digits.
• A starting number you control, so volume two of a production can continue from where volume one ended.
• A skip-first-page option, because covers and title pages usually stay clean.
• Size, margin, and color controls, so the stamp reads clearly without shouting over the document.
• Instant preview before you commit, because finding a collision with existing footers after downloading is nobody's favorite moment.

Notice what is absent from that list: accounts, uploads, email addresses, and per-document fees. The tool is a stamp, not a subscription. You bring the document; it returns the same document with ink on it.

The sequence below covers the common case — a filing-ready exhibit with 'page X of Y' at the bottom. The same rhythm applies to every other convention; only the settings change.

1. 1

   Load the PDF

   Drag your file in or pick it from disk. It opens in the browser and stays there — the preview renders locally, page by page.

2. 2

   Pick the position

   Bottom center is the court's default; bottom right is the academic default; top right earns its keep on documents with busy footers. Flip through a few pages to check for collisions with existing footer text.

3. 3

   Choose the format

   Plain number, 'Page X', 'X of Y', Roman, or Bates prefix with padding. For filings, 'X of Y' is the friendliest — it tells the reader the document is complete in one glance.

4. 4

   Set the range rules

   Decide whether page 1 is the first page or the first page after a skipped cover, and set a custom start number if you are continuing another volume.

5. 5

   Stamp and inspect

   Apply the numbering and flip through the result — especially landscape pages, oversized pages, and any page whose footer already contains text.

6. 6

   Download the numbered file

   Save the output. The original file on disk is untouched, so you can always start over from the clean version.

Every page in a PDF carries a content stream — the page-description program that draws its text and graphics. When this tool numbers a page, it appends text-drawing operators to that stream: set the font, set the position, show the glyphs. Viewers execute those operators after the existing ones, so the number paints on top of the page, in exactly the way a stamp would land on paper.

The engine is pdf-lib, a JavaScript PDF library that parses your document, walks its page tree, computes coordinates from each page's media box — a US Letter page is 612 by 792 points, so 'bottom center' means an x-coordinate near 306 and a y-coordinate a comfortable margin above 0 — and writes the number into the stream. Pages of different sizes get their own coordinates; nothing is stamped blind from a template.

Fonts are where quiet engineering shows. The PDF standard ships with fourteen built-in typefaces — Helvetica among them — that every compliant viewer must render. Drawing numbers in a standard font keeps the output lean, because no font program needs embedding, and guarantees identical rendering in Acrobat, Preview, Chrome, and the clerk's ancient desktop viewer. Custom fonts are available when a filing demands them, at the cost of a slightly heavier file.

One subtlety worth knowing: the PDF format also defines page labels — an optional structure in the document catalog that tells viewers to display pages as 'i, ii, iii, 1, 2, 3' in the sidebar. Those are metadata, invisible on the printed page. Drawn stamps are ink. Courts, binders, and photocopiers only respect ink, which is why this tool draws rather than labels. If you have ever opened a PDF whose sidebar counted differently from its printed footers, you have met the distinction.

Because the whole operation is arithmetic and text on top of existing content, it is also fast. A few hundred pages number in seconds, and a four-figure production set is entirely feasible on a normal laptop — the constraint is your machine's memory, not a server's patience.

Conventions are easiest to see in examples, so here are the four jobs that account for most of the numbering traffic — each with the settings that get it right the first time.

Five jobs, five convention switches, one tool. The pattern across all of them is that the numbering decision is really an audience decision made before the first stamp: who will cite these pages, and in what dialect do they cite them? Answer that, and the settings choose themselves.

For a simple operation, page numbering collects an odd number of failure modes. All of them are avoidable, and all of them are worth knowing before you stamp a 400-page binder.

• Existing numbers. If the document already prints 'Page 4' in the footer, stamping '4' two millimeters away produces a document that reads like a misprint. Choose a different position, or skip numbering entirely.
• Mixed page sizes. Letter and legal pages interleaved in one scan — common in older case files — shift the center line. Check the widest and narrowest pages in the preview, not just the first three.
• Landscape rotation. Pages scanned sideways put the 'bottom' margin where the side is. The number lands technically correct and visually wrong. Rotate first with PDF Rotate, then number.
• Dark or busy margins. A black stamp on a scan with a photocopier shadow stripe disappears. The color and position controls exist for exactly this; top of page beats bottom of page when the bottom is soup.
• Over-numbering. Not every page wants a number — covers, divider tabs, and blank separator sheets are traditionally skipped. The skip-first-page rule handles the cover; for sparse skips, split the document with PDF Split, number the pieces, and merge them back.
• Bates collisions. In litigation, the prefix must be unique across the entire production, including prior productions in the same matter. Before stamping 12,000 pages, confirm what the last production ended at — the start-number setting is how you continue, not overlap.

The meta-rule behind all six: preview before you commit, on pages other than the first three. Numbering defects have a gift for hiding on page 217.

Litigation support staff are the intensity peak. Discovery productions customarily Bates-stamp every page that leaves the firm, because citations in briefs and depositions depend on each page having one immutable identifier. When a partner says 'pull the pages around SMITH-004612,' that only works if the stamping was done once and done right.

Graduate students and journal authors number for format compliance — theses especially, where the registrar's formatting guide specifies Roman front matter, Arabic body, and a specific corner placement, and where non-compliance means a rejected submission rather than a deduction. Paralegals and solo attorneys number exhibits and appendices for filings. HR departments number policy acknowledgments and handbook versions so a signed page can be cited precisely in a dispute years later.

Print brokers and office administrators round it out: anything destined for a binder, a saddle-stitch, or a client meeting table wants numbers, because a dropped stack of unnumbered pages is just confetti with formatting. The common thread is citation — every one of these people needs to point at page 41 and be understood.

Pagination was born from paper's central weakness: loose pages shuffle. A binder, a court file, a deposition transcript — each is a stack of separate sheets pretending to be a book, and numbering is the spell that makes the pretense hold. Remove the numbers and the stack reverts to what it physically is: fifty interchangeable rectangles.

Digital documents were supposed to obsolete all this. Search replaced thumbing; scrolling replaced flipping; a viewer's own page counter sits in the chrome of every PDF reader on earth. Yet the requests kept coming, because citation practice never moved. When a judge says page 41, a professor says see page 118, or a discovery response demands sequential numbering across thousands of produced pages, they are asking for an identifier that survives printing, photocopying, screenshotting, and forwarding — contexts where the viewer's chrome is nowhere to be found.

Numbers also let a document testify about its own completeness. 'X of Y' turns a missing page into an immediately visible accusation; in discovery disputes, a gap in a Bates sequence is a motion waiting to be filed, and in audit work a broken sequence is a finding. A numbered document can be challenged, checked, and reassembled by anyone holding a subset of its pages. An unnumbered one can only be believed.

Official documents reward a sixty-second pre-flight. Every item below exists because somebody, somewhere, learned it the expensive way.

• Match the rulebook. Courts, registrars, and formatting guides specify position and format in writing; match their wording exactly rather than approximately.
• Count pages before and after. The output should be the input count plus nothing — numbering adds marks, not pages. A changed count means the wrong file went in or came out.
• Confirm the skip rule. If page 1 of the output is physically page 2, that was a decision, not an accident. Make it once, deliberately.
• Tour the odd pages. Landscape sheets, legal-size scans, and pages with existing footers are where stamps misbehave; visit at least one of each in the preview.
• Check Bates continuation against the last production. The new sequence must begin after the old one ended, never overlapping it.
• Preserve the clean original. The numbered file is the working copy; the unnumbered file is the source of truth for the next revision.

Then read one numbered page aloud the way a stranger would: 'Exhibit B, page 17 of 86.' If the sentence sounds like it refers to a real, findable thing, the stamping is done.

The table reads as style advice but it is really audience advice. A clerk wants to verify completeness at a glance, so 'of Y' serves them. A litigator wants unique citability, so Bates serves them. A registrar wants structural correctness, so Roman-then-Arabic serves them. The number itself is identical ink in every case; the surrounding convention is what the reader actually reads.

Page numbering has a particular privacy wrinkle: the documents that most need it are the documents least suited to uploading. Discovery productions are confidential by definition. Medical records get compiled into numbered exhibits. HR files, financial statements, and sworn statements all pass through this tool in volume.

So the tool does the work locally. pdf-lib parses your file in the browser, the stamping is pure computation on your machine, and the numbered result is assembled in memory and saved straight to disk. No copy of your exhibit travels anywhere — there is no 'processing server' to travel to. You can disconnect from the network the moment the page loads and stamp a thousand pages on a plane.

The same rule holds across this PDF suite, and it is worth checking for in any tool you point at sensitive documents: Watermark PDF for confidentiality stamps, PDF Merge for assembling exhibits, and the splitter for carving them back apart. None of them uploads. Confidentiality markings that require uploading the confidential document to a stranger's server have always struck us as a strange bargain.

The questions below arrive in one form or another with every batch of newly numbered documents. The answers are short on ceremony and long on practicality.

Numbering is usually the last step of an assembly line, and the earlier steps live nearby. PDF Merge builds the exhibit or thesis from pieces before it gets numbered. PDF Split separates sections that need different numbering conventions — the Roman front matter from the Arabic body. And Watermark PDF adds the confidentiality layer that litigation and HR documents often carry alongside their numbers. Run them in any order; run them all offline; skip the accounts everywhere.