How to reduce PDF file size without losing quality

A PDF that will not fit under an upload limit is one of the more common file problems, and the usual advice — "compress it" — skips the step that determines whether compression will help at all.

PDFs get large for a small number of specific reasons, and the right fix depends entirely on which one applies. Some fixes are lossless and free. Some discard image data permanently. Working out which you need takes about a minute and saves a lot of trial and error.

First, find out what is taking the space

Divide the file size by the page count. That single number tells you most of what you need to know.

Under 100KB per page — the document is mostly text and is already efficient. Compression will achieve very little, because there is nothing large to compress. Look at page count instead: you may be able to send fewer pages.

100KB to 1MB per page — probably a mix of text and images, or a greyscale scan. There is room to improve.

Over 1MB per page — almost certainly a scanned document or one full of high-resolution photographs. This is where compression achieves dramatic results, often 80–90%.

The reason this matters: a twenty-page text report at 400KB total is already about as small as it will get, and no amount of compression will change that. A twenty-page colour scan at 60MB can usually be reduced to under 5MB with no meaningful loss of legibility.

Why scanned documents are so large

A scan is not text. It is a photograph of a page, and it is stored as one.

A single page scanned in colour at 600 DPI is roughly 5,100 × 6,600 pixels — about 34 megapixels, larger than most camera photos. Twenty of those in one document is a genuinely enormous amount of image data, all to represent text that would occupy a few kilobytes if it were stored as characters.

This is also why searching a scanned PDF finds nothing. There is no text in the file, only pictures of text. Making it searchable requires optical character recognition, which is a separate operation from compression.

The lossless fixes: try these first

These reduce file size without discarding anything. Always worth doing before anything lossy.

Remove pages you do not need. The most effective and most overlooked. If a landlord needs two pages of a forty-page bank statement, sending two pages is a 95% reduction, takes ten seconds, and shares far less of your information. Use split PDF to extract exactly what you need — page order is respected, so you can reorder while extracting.

Check for duplicated content. Documents assembled from multiple sources sometimes embed the same logo or image separately on every page rather than referencing it once. Little you can do about it directly, but re-exporting through a different tool sometimes consolidates it.

Subset embedded fonts. A PDF embeds the fonts it uses so it renders identically everywhere. Embedding a full font family can add hundreds of kilobytes; subsetting stores only the characters actually used. Most modern exporters do this by default, but older ones and some office software do not. Re-exporting from a current application often fixes it.

Flatten form fields. Interactive forms carry structure that persists after the form is filled. If nobody needs to edit it again, flattening — printing to PDF is the crude version — converts everything to plain page content and can shrink the file noticeably.

The lossy fix: image compression

If the document is image-heavy, this is where the real reduction lives — and it is genuinely irreversible, so keep the original.

Compression works on a PDF's images in two ways: reducing their resolution and increasing their compression ratio. Both discard data permanently.

The resolution question is the one worth thinking about, because most scans are captured at far higher resolution than they need.

Dropping a 600 DPI scan to 150 DPI removes roughly 94% of the pixels. The text remains entirely readable on screen. That single change is usually the difference between a file that will not send and one that will.

Be more careful with documents containing signatures, small print, stamps, or anything that might need to be examined closely later. Compress conservatively and check the result at full zoom before discarding the original.

The approach that usually works best

Counterintuitively, the most reliable way to shrink a large scanned PDF is often to take it apart and rebuild it.

1. Export the pages as images.
2. Compress the images — this is where the reduction happens, and image compressors give you far more control than PDF compressors typically expose.
3. Reassemble them into a PDF.

This works better than generic PDF compression because you can see exactly what each page looks like at each quality setting, rather than accepting whatever a single "compress" button decides. It also lets you treat pages differently — compressing a page of plain text harder than one containing a signature.

The trade-off is that any text layer is lost, so a searchable PDF becomes non-searchable. For a scan that was never searchable anyway, there is nothing to lose.

Order matters

If you are doing several of these, the sequence affects the result.

Remove pages first. No point compressing pages you are about to delete.

Compress once, at the end. Compressing several files and then merging produces a worse result than merging and compressing once, because each lossy pass degrades the images further. If you are combining documents with merge PDF, merge first.

Never compress an already-compressed file. The second pass treats the first pass's artefacts as real detail and preserves them while discarding genuine detail. Always work from the original.

Stop the problem at the source

If you scan documents regularly, the scanner settings matter more than anything you do afterwards.

Scan in greyscale, not colour, unless colour carries meaning. A typed document scanned in greyscale is a third the size with no loss of legibility.

Use 300 DPI, not 600. Adequate for both screen and print. Reserve 600 for photographs and documents with genuinely fine detail.

Use black-and-white mode for plain text. Some scanners offer a one-bit mode that produces astonishingly small files for typed documents. It looks harsh and is unsuitable for anything with photographs or shading, but for a printed letter it is ideal.

Turn on the scanner's own OCR if it has it. The file stays roughly the same size, becomes searchable, and you avoid needing a separate step later.

Getting these right means most documents never become a problem in the first place.

What compression will not fix

Text-only PDFs are already small. If the file is 200KB and mostly text, there is nothing meaningful to compress. Reduce the page count or accept the size.

Vector graphics do not compress like images. Charts, diagrams and CAD exports are stored as drawing instructions rather than pixels. A complex vector illustration can be large, and image compression does nothing to it — the fix is simplifying the artwork in its source application.

Embedded files stay embedded. PDFs can carry attachments — spreadsheets, other PDFs — that contribute their full size and are invisible in normal viewing. If a file is inexplicably large, check the attachments panel.

A note on the tools

Every PDF tool on this site processes files in your browser rather than uploading them. For this particular task that is more than a preference: the documents people compress are frequently bank statements, medical letters, signed contracts and identity scans, and the conventional way to shrink them involves sending the full-resolution original to a server first.

The split, merge and JPG to PDF tools all run locally, with no file size limit imposed by us — because there is no bandwidth for us to pay for.