A PDF never gets large by accident. Something inside it is carrying weight the document doesn't need, and in practice that something is almost always one of four things: images stored bigger than they appear, scanned pages, duplicated content, or metadata and structural overhead. Each cause has a different fix — and applying the wrong one is how people compress a text-only document three times and wonder why nothing shrank.
This is a diagnosis guide. If you first want to know whether your file's size is even abnormal for its kind, PDF File Size Guide covers what's normal and what's too big, and How to Compress a PDF for Email covers the step-by-step when an attachment limit is your immediate problem. This post goes the other direction: into the file, cause by cause, with the fix that matches each one.
First, work out which cause you have
You can usually narrow it down in under a minute with three questions. Did the file come from a scanner or a camera app, or was it exported from a program? Does the document contain photos, diagrams, or charts, or is it text on plain pages? Was it assembled from several sources — merged files, repeated appendices, copies of copies?
A scan answers the question on its own: every page is a photograph, and the weight is real. A digital export with images inside points at embedded-image size. A merged document assembled over months points at duplication. And a plain typed document that's still somehow huge is the rare case where structure and metadata are the whole story. Everything below follows from that triage.
Cause 1: Images stored bigger than they appear
This is the most common cause by a wide margin. A photo placed two inches wide in the final layout doesn't need four thousand pixels of source data behind it — but many export tools embed the full-resolution original anyway. The document looks identical on screen either way, which is exactly why the mismatch survives until an email bounce exposes it.
The fix is compression that re-encodes images to the size they actually display at. At the Medium and High levels, Compress PDF detects each image's true on-page size and re-encodes it to your chosen DPI and JPEG quality — shrinking the stored data to match the displayed reality. Low is the wrong medicine here: it only strips metadata and optimizes structure, leaving images untouched.
The honest trade-off: this re-encoding is real. It's invisible at reading size but visible if you zoom in — so keep the original if the document's images will ever be cropped, printed large, or inspected closely.
Cause 2: Every page is a scan
A scanned page is a full-page photograph, so a scanned document is heavy because its content genuinely is heavy. You can compress scans — they're images, and Medium or High will re-encode them like any other — but the better fix happens before the PDF exists, at the scanner.
Three choices dominate the outcome. Resolution: 200–300 DPI is the standard range for readable text, and doubling DPI roughly quadruples the pixels per page. Color mode: a color scan of a black-and-white page stores three channels where one would do — grayscale keeps the tone and detail at a fraction of the weight. Coverage: blank sheets and double-fed pages end up in the file just like real ones, and Remove Pages can strip those after the fact, with the remaining pages copied across exactly.
If you scan with a phone, Scan to PDF builds the file with an enhancement mode — including grayscale and black-and-white — chosen at capture time, so the document is born closer to the right size.
Cause 3: Duplicated and redundant content
The quietest cause. A file merged from several versions of itself, a forty-page appendix that appears twice, a report where the same chart sits in three sections — none of it looks wrong on screen, and all of it weighs what it weighs.
When the duplication is whole pages, the fix is direct: remove them. Remove Pages takes the pages and ranges you list — 41-80, say — and keeps every other page in its original order, at full quality. When the duplication lives inside the document's resources — the same logo stored separately behind every page — the honest answer is that compression re-encodes images to their display size, which shrinks each copy, but it is not a de-duplication tool. The reliable fix there is re-exporting from the source file with the repetition cleaned up.
Cause 4: Metadata and file structure
The minor one, and worth saying plainly: metadata and structure overhead is usually measured in kilobytes, not megabytes. Document properties, edit history, unused objects left behind by earlier edits — real, but small. If a document is huge and contains nothing but text, the cause is almost never the metadata; it's that the document is long.
The fix is the Low compression level: it strips metadata and optimizes the file's structure while leaving every image untouched — no quality trade-off at all. That makes Low the right choice for text documents, where there's nothing to re-encode and nothing to gain from trying.
The cause-to-fix cheat sheet
- Digital export with images — compress at Medium or High, which re-encode images to their display size. Keep the original.
- Scanned document — rescan at 200–300 DPI in grayscale for text pages if you can; otherwise compress at Medium or High.
- Repeated pages or merged-in duplicates — remove the duplicate pages; the survivors keep full quality.
- Text-only but still big — Low tidies metadata and structure with zero quality impact, and the file is probably just long.
What compression honestly won't fix
Two limits are worth stating before you press the button. First, a document that's large because it legitimately contains a lot can't be made dramatically smaller — text is already tiny, and shrinking images only helps when images are the weight. Second, a detail that surprises people: if the PDF is password-protected, compressing it rebuilds the file without that protection, so keep a copy of the original if the password matters.
Diagnose first, then compress with the level that matches the cause — that's the whole method.