How to Reduce a Scanned PDF's File Size Without Making It Unreadable

The situation

A 40-page lease scanned at the office copier comes out at 95 MB. The same lease as a Word export would be 400 KB. Nothing is wrong with the scanner; a scan simply stores every page as a photograph, and photographs are expensive.

Shrinking a scan safely means deciding how much photograph each page really needs.

Why scans are so large

Every scanned page is a bitmap. A letter-size page at 300 dpi is about 2550 × 3300 pixels — roughly 8.4 million pixels. In color that is 25 MB uncompressed. JPEG brings it to 0.5–3 MB depending on quality, and 40 pages of that is 20–120 MB.

Three settings decide the number:

1. Resolution (dpi). Pixel count grows with the square of resolution. 600 dpi is four times the data of 300 dpi; 150 dpi is one quarter.
2. Color depth. Color stores three channels, grayscale one, black-and-white (bilevel) one bit per pixel. Going from color to grayscale alone often cuts a scan by 40–60%.
3. Compression method. Photos and grayscale pages use JPEG (quality-adjustable). Bilevel pages use CCITT Group 4 or JBIG2, which are far smaller for clean typed text — a 300 dpi bilevel page is often 30–100 KB.

Typical per-page sizes, for orientation:

What legibility actually requires

• Typed text, 10–12 pt: readable at 150 dpi; comfortable at 200 dpi.
• Small print, footnotes, 6–8 pt: keep 200 dpi or more.
• OCR (making the text searchable): 300 dpi is the standard input; OCR accuracy drops noticeably below that.
• Handwriting, signatures, stamps, pencil notes: keep grayscale. Bilevel conversion drops faint strokes and turns red stamps into black blobs or nothing.
• Numbers that matter (invoices, drawings, lab results): avoid lossy JBIG2. In 2013, Xerox scanners using aggressive JBIG2 settings were found to substitute digits — a 6 became an 8 — in scanned documents. Lossless modes do not have this problem.

Reviewers at courts, agencies and clients read on screen at 100% zoom, sometimes on a laptop. Design for that, not for a magnifying glass.

Checklist

• Do you still have the paper? Re-scanning at the right settings beats any compression.
• Which pages carry stamps, seals, signatures or color-coded marks? Those set the floor.
• Does the file have an OCR text layer? Check with Ctrl+F for a word on page 1. You will want to confirm it survives.
• What is the target? A number, not "smaller".

Step by step

1. If you can re-scan: choose grayscale, 200–300 dpi, "text" or "document" mode, PDF output. Most copiers have a compact PDF or high compression option that applies exactly this.
2. If you only have the file: open PDF300, enter the target size, and download. Target-size compression chooses the resolution and quality for you instead of making you guess between "medium" and "strong".
3. Check three pages at 200% zoom: the first, one dense page in the middle, and the one with the smallest print or a stamp.
4. Confirm the OCR layer with Ctrl+F. Some tools re-rasterize pages and drop the invisible text; if yours did, run OCR again on the compressed file or use a tool that keeps it.
5. Still too large? Split the scan into logical parts — exhibits, chapters, date ranges. Each part keeps full legibility.

FAQ

Grayscale or black-and-white?

Grayscale for anything with stamps, handwriting, signatures, highlighter or photos. Black-and-white only for clean, typed pages — and even then, check that thin fonts and underlines survived.

My 50 MB scan only dropped to 45 MB. Why?

The scan was already low resolution with heavy JPEG compression, and the remaining size is page count times the minimum cost per page. The lever left is grayscale (if it is color) or splitting.

Does compressing a scan remove the text I can search?

It depends on the tool. The OCR layer is small and most tools preserve it, but some rebuild pages as plain images. Test with Ctrl+F after compression; it takes five seconds and catches the problem before you send the file.