Never leaves your device
Your browser reads, rewrites and saves the file locally. There is no upload step, no server copy to delete afterwards, and no account to create first. Close the tab and nothing of the document remains anywhere.
Shrink a PDF in your browser, by level or target size.
Drop your PDFs here. They are compressed on your device — nothing is uploaded.
Compression runs in two passes. Photographs are decoded, scaled down when they are larger than the page needs, and re-encoded at the quality you chose. Then the structure is rewritten, so cross-reference bookkeeping packs down and unreferenced objects are dropped. Text, vectors, bookmarks, form fields and tagged structure all come through intact, and it happens in your browser tab.
Your browser reads, rewrites and saves the file locally. There is no upload step, no server copy to delete afterwards, and no account to create first. Close the tab and nothing of the document remains anywhere.
Light rewrites structure only and leaves every pixel alone. Medium and Heavy re-encode photographs, trading visible detail for size. Or type a size — 2 MB for a job portal, 5 MB for email — and the tool works towards it and reports honestly whether it got there.
Outlines, form fields, internal links, page labels and tagged-PDF structure are carried through rather than flattened away. That last one is what screen readers rely on, and it is the first thing most compressors quietly discard.
Drop one file or twenty onto the page, or pick them from your device. Nothing is sent anywhere while they load; the page reads each document directly.
Choose Light, Medium or Heavy depending on whether quality or size matters more, or switch to target size and name the number you have to hit.
Each file reports what it started at, what it ended at, and how much came off. Open the comparison to see a page before and after, then save one file or take the whole set as a ZIP.
It depends almost entirely on what is inside. A file full of photographs can drop by half or more at Medium, because the images are where the bytes are. A text report has little to give and may only lose the structural slack, which is often a few percent. Nobody can promise a figure in advance, which is why the result shows you the real one.
Light rewrites the file's structure and touches no image data at all, so pages come out visually identical — use it for print and archiving. Medium re-encodes photographs and scales oversized ones down, which suits email and everyday sharing. Heavy pushes both further for the smallest practical file, and is the one worth checking before you send it.
You name a size and the tool works towards it, measuring the real output each time rather than guessing. It is a goal, not a guarantee: a PDF made mostly of text has nothing left to give at any quality, and no setting will drag it under an arbitrary number. When that happens the result says the target was not reached and suggests splitting the document instead. Split PDF
At Light, no — nothing is re-encoded and the pages are unchanged. At Medium and Heavy the photographs are re-encoded, so they are genuinely lossy, and large ones are scaled down. Text and vector drawings are never re-encoded at any level, so they stay sharp however small the file gets.
Usually yes, if the scan is made of photographs — that is exactly the case Medium and Heavy are built for, and it is often where the biggest reductions come from. A black-and-white, fax-style scan is stored in a format this tool leaves alone, so it will barely move; rescanning it at a lower resolution gets much further. Scan Document
Yes, that is most of what Medium and Heavy do. Photographs are decoded, scaled down when they are larger than the page needs, and re-encoded. Some images are deliberately skipped because rewriting them would break the page: transparency masks, palette-based artwork, print-ready CMYK and stencil images are all left exactly as they were.
Yes. The outline, form fields, internal links, page labels and the tagged structure that screen readers use are all carried across. That is not a given elsewhere: many compressors rebuild the document from its pages alone, which silently throws every one of those away along with the document's title and author.
Because there was nothing left to remove. The file had no structural slack and its images were already efficient or of a kind this tool skips. When that happens you get your original bytes back and the page says so, rather than counting a same-size copy as a saving. Your document is not damaged and nothing was lost.
Switch to target size, type the limit you are working to, and let it aim for that. If the file still will not fit, the honest answer is usually that it is one document too many: splitting it into parts, or exporting the pages you actually need, gets you under a hard cap when re-encoding cannot.
No. The work happens in your browser tab, so the document never travels over the network and nothing is left sitting on a server afterwards. There is no watermark, no account, and no copy of your file anywhere but on your own device.
Try this tool with a sample, or upgrade to use your own files.
$9.99 per month
Secure checkout. Cancel anytime.
Enter the email address you used to purchase PDF Head Pro. We will send a secure, single-use link.
Secure checkout. Cancel anytime.
If that email has eligible access, we sent a verification link.