Compress PDF without losing quality: what actually works

---

A PDF that was 40 MB an hour ago is still 40 MB after you have tried "save as" three different ways. Email rejects it, the upload form caps out at 25 MB, and every guide promises you can compress PDF without losing quality. What none of them explain is what compression actually does to the file, which is the part that tells you whether you will like the result.

What you can compress PDF without losing quality on, and what you cannot

Almost never the text. A page of body text, even a long one, is a few kilobytes — character codes and font references, not much else.

The weight comes from two places:

Images, especially scanned pages. A single page scanned at 300 DPI in colour can be several megabytes on its own, and a scanned document is a stack of those. This is by far the most common cause of an oversized PDF.

Embedded fonts. A PDF that embeds a full font file — every glyph, every weight — so it displays identically on a machine that does not have that font installed. Do that with two or three typefaces and it adds up, though rarely as much as images do.

Duplicated resources are the quieter third cause: the same image or font embedded separately on ten different pages instead of once and referenced ten times, which happens more often than it should when a document has been assembled from other documents.

Lossless vs lossy, and why the distinction matters

Lossless compression removes waste without touching content: deduplicating an image embedded multiple times, stripping unused fonts or metadata, re-encoding data more efficiently. The page looks and reads exactly as it did. The size reduction is real but modest — usually a small percentage on a document that was already reasonably built.

Lossy compression changes the content to make it smaller. For a PDF this almost always means the images inside it: reducing resolution (downsampling), and increasing JPEG compression (which discards fine detail the eye is least likely to miss, until you push it too far).

Guides that promise "compress without losing quality" are describing the lossless case, or a lossy pass set conservatively enough that the loss is not visible at normal viewing size. Neither claim survives being pushed hard — there is no setting that turns a 40 MB scanned document into 2 MB with nothing given up. The size and the fidelity are trading against each other; compression only decides where the exchange rate is set.

The resolution most people never think about

The DPI a scan actually needs is lower than the DPI it is usually captured at, and the gap is most of the wasted size.

A page you are going to read on a screen needs roughly 150 DPI to look sharp. A page you are going to print needs closer to 300. A page scanned at 600 DPI "to be safe" is carrying four times the pixel data of the 300 DPI version for no visible difference in either use case — the eye and the display both cap out well below that.

This is why downsampling is usually the single biggest lever: it is discarding resolution the document was never going to use, rather than damaging the resolution it needs.

Re-compressing an already-compressed image

Scanned pages are usually stored as JPEG inside the PDF, and JPEG is already a lossy format. Compressing it again does not undo the first pass — it adds a second round of loss on top of the first.

This is why the same compression setting can look fine on one document and visibly blocky on another: it depends on how much headroom the original image had left. A PDF built from a high-quality source photographed once tolerates a second compression pass well. A PDF that has already been through two or three rounds of "shrink it a bit" starts showing artefacts fast, because there was nothing left to give.

How to compress PDF without losing quality, in order

1. Try lossless first. If the tool offers it, run it before anything else — it costs you nothing in quality and sometimes solves the problem outright.
2. Set resolution to the actual target, not the highest number available. Downsampling to 150 DPI for a document nobody will print is not a compromise, it is removing data that was never doing anything.
3. Compress images moderately rather than aggressively. A moderate pass on every image usually beats a maximum pass on some — the visible damage from over-compression is not linear, it gets much worse quickly past a point.
4. Keep the original. Once a page has been downsampled, the detail is gone — there is no un-shrinking it. Compress a copy, not the only version you have.

If the file is large because it is a scan rather than a document, compression is the wrong lever entirely — see how to make a scanned PDF searchable, which covers what a scan actually contains.

Where compress PDF without losing quality stops being true

"Compress" alone, with no options, on a document that is already efficient. If the PDF is mostly text with one or two reasonable images, there is little left to compress — the tool is not being lazy, there is simply not much waste to remove.

Pushing quality all the way down on a document with fine text or diagrams, then being surprised that small text blurs. That is the lossy trade happening exactly as described — the fix is a less aggressive setting, not a different tool.

Compressing a document more than once, expecting further improvement. Each additional pass on already-lossy content mostly adds artefacts rather than removing size, for the reason above.

The honest expectation for a scan-heavy PDF is a large reduction — often 70–90% — because the original resolution was almost always higher than it needed to be. For a mostly-text PDF, expect a small reduction and treat anything more as a bonus.

If the pages only need to be looked at rather than kept as a document, converting them is often a better answer than compressing them — PDF to JPG on Windows gets a far smaller file than any compression setting will.