A hundred pages of text can be under a megabyte while a three-page scan runs to forty. Understanding where PDF weight comes from makes every size problem easier to solve.
PDF file sizes seem arbitrary until you know what is inside them. A 200-page novel exported from Word might be 900 KB. A three-page scanned form can be 40 MB. The difference is not length — it is what the pages are made of.
Text is almost free
Text in a PDF is stored as characters with positions and font references. A page of dense prose is a few kilobytes. Compressed, often less. This is why digitally created documents — exported from a word processor, generated by a reporting system — are usually small regardless of page count.
Images are the whole story
An image is stored as pixel data, and pixel counts grow with the square of resolution. Doubling the scan DPI quadruples the data.
Rough figures for a single A4 page:
- 150 DPI greyscale — around 1 MB uncompressed, often far less after compression
- 300 DPI greyscale — around 4 MB
- 300 DPI full colour — around 12 MB
- 600 DPI full colour — around 50 MB
Multiply by page count and the arithmetic explains most "why is this so big" questions immediately. A twenty-page colour scan at 600 DPI is a gigabyte-class document before compression.
The scanner settings that cause most damage
- Colour mode on monochrome originals. Scanning a typed black-and-white letter in colour triples the data for zero benefit.
- Archival DPI for everyday documents. 600 DPI exists for reproducing photographs and fine detail. For text, 300 DPI is generous and 200 is usually readable.
- Lossless image formats inside the PDF. Some scanners embed uncompressed or PNG-style data, which is enormous compared with JPEG for photographic content.
If you control the scanner, 300 DPI greyscale is the sweet spot for text documents — small, sharp, and more than adequate for OCR.
The other contributors
- Embedded fonts. Usually modest, but a document using many font families with full character sets can carry several megabytes. Subsetting — embedding only the characters used — avoids most of this.
- Vector graphics. Charts and maps with enormous numbers of paths can be surprisingly heavy.
- Revision history. Some editors append changes rather than rewriting the file, so a heavily edited document accumulates dead data. Rewriting it compacts the file without touching the visible content.
- Embedded attachments and metadata. Less common, but PDFs can carry whole files inside them.
What to do about it
Diagnose before treating. If your file is large and the pages are scanned images, resolution is the lever — recompressing the pages will produce dramatic savings. If the file is large and the text is selectable, images embedded in the document or accumulated revision data are the likely cause, and a lossless structural rewrite may help without any quality loss.
And before compressing at all, check whether you need every page. Extracting the relevant section often solves a size problem completely, and produces a better document for the recipient.
A rule of thumb
If a text document is over about 5 MB, something is wrong — it is probably scanned when it did not need to be, or scanned far too finely. If a scanned document is under 500 KB per page, it is well optimised. Between those, you have room to improve but nothing is broken.
Pixellwork Editorial
Documents & Privacy
Published October 6, 2026