How to compress a PDF
- Drop one or more PDF files into the box above, or click it to choose them.
- Pick a level: Recommended for most documents, Strong for the smallest file, Light for print.
- Check the new size next to each file. Click Compare to switch between the original and the compressed page.
- Download the compressed PDF, or all of them as a ZIP.
Changing the level recompresses every file straight away, so you can try Strong, look at the result and step back to Recommended in a few seconds. Your original file is never changed — the compressed copy is saved as a separate file, so you keep both.
What the compression levels do
The level decides one thing: how much detail images keep. Text, fonts and vector drawings are identical at every level.
| Level | What happens to images | Best for |
|---|---|---|
| Light | Above 450 DPI → 300 DPI, high JPEG quality | Brochures, photo books, anything you will print |
| Recommended | Above 225 DPI → 150 DPI, good JPEG quality | Invoices, reports, scanned forms, slides |
| Strong | Above 130 DPI → 100 DPI, lower JPEG quality | Email attachments and upload forms with a size limit |
DPI here means the resolution of an image as it is printed on the page, not its pixel count. A 3,000-pixel-wide photo placed four inches wide on a page is 750 DPI — far more than any screen or office printer can show. At Recommended it becomes 600 pixels wide, which is still 150 DPI at that size, and usually a fraction of the original bytes.
Why is my PDF so big?
Almost always, images. A page of text takes a few kilobytes, because letters are stored as characters. A scanned page or a phone photo is stored as pixels, and scanners and phones capture far more of them than a document needs. Three other causes are common enough to check for:
- Hidden editing data. Illustrator and Photoshop can save their whole working file inside the PDF so it can be edited again. The page looks the same, but the file can be several times larger.
- The same thing stored many times. Merging PDFs from one template often repeats the logo and the fonts once for every document.
- Leftovers. Deleted pages and earlier versions of an edited file can stay inside it, invisible but still counted.
After compressing, open What was taking up space under a file. It splits the file into images, fonts, text and drawings, hidden editing data and file structure, before and after — so you can see whether a stubborn file is image-heavy or simply full of fonts.
Why the text stays sharp and selectable
A common shortcut in online PDF compressors, especially ones that run in the browser, is to turn every page into a single picture and save it as a smaller JPEG. It is easy to build and it does make files smaller, but the text becomes a blurry image: you can no longer select it, search it, copy it or have it read aloud, and links stop working.
This tool works on the parts of the PDF instead of on the finished page. It reads each page's drawing instructions to find how large every image is actually printed, including images inside reused page elements and form fields, and resizes only images that are sharper than the level needs. Everything else is written back as it was: real text, embedded fonts, vector graphics, links, bookmarks, form fields and the invisible text layer that OCR adds to scans. Drawings and logos with sharp edges are not converted to JPEG, because JPEG smears hard edges; they are only packed tighter without any loss.
What is removed, and what is kept
| Removed or reduced | Kept exactly |
|---|---|
| Oversized photos and scans (resized to the level's DPI) | Text, fonts and vector drawings |
| Hidden editing data from Illustrator and Photoshop | Links, bookmarks and form fields |
| Page thumbnails and per-image metadata | The OCR text layer of scanned documents |
| Duplicate copies of the same image or font | Title, author and document metadata |
| Deleted pages and objects nothing refers to | Attachments embedded in the PDF |
| Uncompressed or weakly compressed data (packed losslessly) | Page size, rotation and order |
If a PDF contains a digital signature, the tool tells you before you download: any change to a signed file, including compression, makes the signature show as invalid.
When a PDF will not shrink much
Some files are already about as small as they can be without damaging them. Text-only documents exported from Word or Google Docs typically shrink by only a few percent, if at all. Images already at a sensible resolution are left alone, because saving a JPEG again at lower quality costs detail for very little gain. Scans stored as JPEG 2000, CCITT or JBIG2 — compact formats that many office scanners use — are kept as they are, and so are images in CMYK, the colour format of print-ready files, because converting them in a browser would shift their colours. When the compressed copy would not be smaller than your original, the tool says so and does not offer a larger file.
Getting under an email or upload limit
Email services usually cap attachments at around 20–25 MB, and application portals, visa and university forms often allow only 1–5 MB per document. Start with Recommended; if the file is still too large, try Strong and use Compare to check that the pages are still readable. For a long scanned document that is still over the limit, split it into parts with Split PDF or remove pages nobody needs with Delete PDF Pages, then compress again. Building the PDF yourself from phone photos? Shrinking the photos first with Compress Image to a Target Size and then combining them with JPG to PDF gives you exact control over each page.
Your documents stay on your device
The best-known PDF compressors upload your file to their servers, process it there and delete it later, and several of them keep their stronger compression levels for paying users. Here the whole job runs inside your browser tab, using your own device's processor. Nothing is uploaded, so there is nothing to delete afterwards and no limit on file size, page count or how many PDFs you compress — which matters when the document is a contract, a bank statement, a medical record or an ID scan.