Most people assume a PDF has a page size — that "this is an A4 document" is a property of the file. It isn't. A PDF stores a size on every individual page, and nothing forces those sizes to agree. A single file can legitimately hold an A4 page, a Letter page, a landscape exhibit, and a slightly-off scan, all in sequence. Usually they happen to match because they came from one source. When they don't, you get a document that looks fine on screen and falls apart the moment it's printed, bound, or fed to a system that expects uniformity.
This guide explains where page size actually lives inside a PDF, why documents end up with mismatched pages, and what "normalizing" them really does.
Page size is a per-page property
Inside a PDF, each page is its own object, and one of its attributes is a rectangle that defines the page's physical size. The whole file has no single "document size" field. So when you open a 50-page PDF, you're really looking at 50 independent size declarations that happen, most of the time, to be identical.
This design is deliberate. PDF was built to represent anything a printer could output, including documents that genuinely mix sizes: a report with a fold-out chart, a legal filing with oversized exhibits, a book with differently-trimmed inserts. The format never assumed uniformity, so it never enforced it.
The practical consequence: uniformity is something a document has by luck or by effort, not by default.
The boxes that define a page
A PDF page can carry up to five different rectangles, called page boxes. Two of them matter for everyday work:
- MediaBox — the full physical extent of the page. This is the "paper size." Every page has one; it's the closest thing to "the page size."
- CropBox — the visible region. Viewers and printers clip the page to this rectangle. If the CropBox is smaller than the MediaBox, the page displays smaller than its actual paper size, with the rest hidden but still present.
The other three (BleedBox, TrimBox, ArtBox) are print-production boundaries that describe where the paper gets trimmed and where bleed extends. They matter to commercial printers; they rarely matter for office documents.
The MediaBox/CropBox distinction is why two operations that look similar are actually different:
- Cropping changes the CropBox — it hides part of the page without changing the underlying paper. Crop PDF works here.
- Resizing changes the actual page dimensions and scales the content to fit. Resize PDF Pages works here.
A cropped page can still report a large MediaBox; a resized page genuinely is a different size.
Why pages drift apart
Mismatched page sizes almost always come from one of these:
Scanning. Scanners that auto-detect document edges produce a MediaBox sized to whatever they measured — a few millimetres off true A4, and slightly different from page to page. A 30-page scanned contract can have 30 subtly different sizes, none of them exactly standard.
Merging from mixed sources. When you merge PDFs, each source keeps its own page sizes. Merge an A4 report with a Letter-sized appendix and the combined file alternates between the two. The merge operation copies pages faithfully; faithful copying means preserving their original sizes.
Mixed regional templates. A4 is the global standard; Letter is North American. A document touched by people on both sides of that divide picks up both sizes. Each contributor's pages look correct in isolation.
Generated or exported pages. A spreadsheet exported wide, a slide deck at 16:9, a cover page built from an image — each lands at whatever size its source produced, which may not match the body.
A single rogue page. Often it's just one: an inserted exhibit, a fax cover sheet, a wrongly-templated signature page. One outlier in an otherwise uniform document.
Why it matters
If it only ever lived on a screen, mismatched sizes wouldn't matter — viewers happily render each page at its own size. The problems start the moment the document leaves the screen:
- Printing. A printer set to one paper size has to scale or shift the odd pages, producing inconsistent margins, off-center content, or unexpected scaling. Double-sided printing is worse: a single odd page throws off front/back alignment for everything after it.
- Binding and stapling. Physical assembly assumes a consistent trim. One wider page sticks out of the stack.
- Strict intake systems. Court e-filing portals, regulatory uploaders, and some document-management systems reject files whose pages aren't all a required size (commonly Letter or A4). They check the MediaBox and refuse anything outside spec.
- Downstream processing. Stamping page numbers, applying watermarks, or imposing pages for a booklet all assume a predictable page size; outliers land in the wrong place.
What "normalizing" actually does
Normalizing means rewriting the odd pages' dimensions so every page agrees on one size. The mechanical question is what happens to the content that was sized for the old dimensions, and there are three honest answers:
- Fit with padding keeps the content at its proportions and adds whitespace to fill the new, larger box (or scales down to fit a smaller one). Nothing distorts, nothing is lost. This is almost always the right choice.
- Stretch to fit scales width and height independently to fill the target exactly. It wastes no space but can distort content, since the two axes scale by different amounts unless the aspect ratios already match.
- Crop to fit scales content to cover the new page and trims the overflow. No distortion, but edge content can be cut.
The size of the target is a separate decision from how content fits into it. You might target the document's existing majority size (so the few outliers fall in line) or force a hard standard like A4 or Letter (so the whole document conforms regardless of what it was). The Resize PDF Pages guide walks through each of those choices in the tool.
Common questions
How do I tell what sizes my PDF's pages are?
Most PDF readers show the current page's dimensions in document properties or a page-info panel. The Resize tool does this for you: it scans the file, reports the majority page size, and shows what fraction of pages match it — so you can see at a glance whether you have a uniform document with a couple of outliers or a genuinely mixed one.
Is "A4" stored in the file, or just the dimensions?
Just the dimensions. A PDF page doesn't record "this is A4" — it records a rectangle (for A4, 595 × 842 points, which is 210 × 297 mm). Software recognizes that those dimensions are A4. This is why a scan that's a millimetre off isn't "A4" to a strict checker even though it looks identical.
Why do dimensions show up in "points"?
PDF measures internally in points: 1 point = 1/72 inch. A4 is 595 × 842 points, Letter is 612 × 792. Tools convert to millimetres or inches for display, but the underlying number is always points. That's also why resize tools offer pt as a custom unit alongside mm and in.
Does changing page size change the PDF version or break anything?
No. Page size lives in each page's dictionary and is independent of the PDF version in the file header. Resizing pages doesn't alter the version, the encryption, or the text layer — it changes dimensions and repositions content.
Can a single page be a different size on purpose?
Absolutely, and sometimes you want exactly that — a fold-out map or a wide data table that's meant to be oversized. Normalizing is only the right move when the variation is accidental. When the odd page is intentional, leave it (or normalize "Only outliers" and exclude the one you want to keep by handling it separately).
Wrap-up
A PDF doesn't have one page size; it has one per page, defined by each page's MediaBox, and nothing forces them to match. Scanning, merging, mixed templates, and stray inserts all produce documents where the sizes disagree — invisible on screen, disruptive the moment you print, bind, or submit the file.
Fixing it means normalizing the pages to a single size and choosing how the existing content fits the new dimensions (padding is the safe default). The Resize PDF Pages guide covers doing it step by step.