PDF File Size Explained: What Makes a PDF Small or Large

A PDF’s size is dominated by what is embedded in it. Scanned pages are images and are the heaviest thing a PDF can contain; embedded font files add 50KB to several hundred KB each; text itself is almost free. A 40-page text document can be 200KB while a single scanned page can be 4MB.

The four things that take up space

Images come first by a wide margin. A page scanned at 300 DPI in colour is roughly 8.7 million pixels before compression. How well it compresses depends on the encoder the scanner chose — JPEG, JPEG2000, or lossless Flate for a black-and-white scan.

Embedded fonts come second. A PDF that embeds a full Unicode font can carry several hundred kilobytes of glyph outlines. Subsetting — embedding only the characters actually used — usually cuts this to a fraction.

Vector graphics are third, and are usually modest: a chart is a list of drawing operations, which compresses well. The exception is a map or a CAD export, where hundreds of thousands of path segments add up.

Metadata and structure are last and normally trivial: XMP metadata, the cross-reference table, bookmarks, form field definitions and digital signatures.

Why identical-looking PDFs differ in size

Two PDFs of the same document can differ tenfold depending on how they were produced. Exporting directly from a word processor embeds real text and subsetted fonts. Printing to PDF and scanning the result produces images of text, with no searchable characters and a much larger file. If your PDF cannot be searched with Ctrl+F, it is a stack of images, which explains its size.

Making a PDF larger on purpose

Portals that set a minimum PDF size are usually trying to reject unreadable scans, the same reasoning behind image minimums. When your PDF is legible and merely efficient, padding it with an inert object gets it over the line without altering a single page. The PDF specification allows objects that nothing references, and readers ignore them, so the document renders identically.

The PDF tool on this site does this in your browser using pdf-lib. The file is parsed, padded and re-saved locally, and never uploaded.