How to compress a PDF
- Drop your PDF on the dashed box, or press Enter on it to choose a file. No file handy? Select Try a sample scan. The line below shows how many images it holds and their share of the size.
- Pick a Compression level: Light, Medium or Maximum.
- Keep Recompress and downsample images ticked, because that is where the savings come from. Tick Remove metadata or Flatten form fields only if you want those changes.
- Select Compress PDF, compare the original and compressed sizes, then select Download compressed PDF.
How the compression works
Each image stored as a JPEG, or as plain RGB or gray pixels, is decoded, resized so its longest side is at most the level's dpi times the longest page side in inches, and saved again as a JPEG at the level's quality. Light is 300 dpi at quality 85, Medium is 150 dpi at 70 and Maximum is 100 dpi at 50. The new image replaces the old one only if it is at least 8% smaller (25% smaller when the original was lossless). Uncompressed data streams are deflated and objects that nothing points to are dropped. Text, fonts, links, bookmarks and form fields are not changed. The formats involved are defined in the PDF 1.7 specification (ISO 32000-1), and the JPEG encoding uses the browser's canvas.toBlob.
Worked example
Select Try a sample scan. It is a one-page US Letter PDF with one color image of 2550 x 3300 px (300 dpi), and the box shows 1 page and 2.88 MB. At Medium the longest allowed side is 150 x 11 = 1,650 px, so the image becomes 1275 x 1650 px and the tool reports "Image 2550 x 3300 px resized to 1275 x 1650 px, 2.88 MB to 174.0 KB." The file goes from 2.88 MB to 174.9 KB, shown as -94.1%. Light keeps the image at 2550 x 3300 px and ends at 2.35 MB (-18.3%), and Maximum resizes it to 850 x 1100 px and ends at 47.5 KB (-98.4%). The sample is synthetic, so another browser may differ by a few percent, and your own scan will vary.
Limits and edge cases
- Left untouched: CMYK images, JPEG 2000, black-and-white fax-style scans, palette images, masks, JPEGs with a rotation flag, and images over 60 megapixels. The result panel counts them.
- Encrypted PDFs are refused. Unlock them first, then compress.
- Changing a file can break a digital signature or PDF/A conformance.
- A PDF that is all text barely shrinks. For single pictures use the Image Compressor; to keep only some pages use the PDF Page Extractor or the PDF Splitter.
- To see what a file currently reveals about you, open it in the PDF Metadata Viewer.
Frequently asked questions
Why didn't my PDF get any smaller?
Text, fonts and vector drawings are already compact, so only images and uncompressed data can shrink. If the line under the upload box says the images take a small share of the file, expect little change. Images that are already heavily compressed, black-and-white scans and CMYK photos are left alone.
Will compressing make my PDF blurry?
Only the images change. At Medium (150 dpi, quality 70) a scan stays readable but fine detail softens a little. Choose Light (300 dpi, quality 85) for small print or photos you plan to print. Text and vector graphics are never touched.
How do I get a PDF under 1 MB or 2 MB for an upload form?
Try Medium first, then Maximum. The tool cannot aim at an exact size, so if the result is still too big, split the file into parts with the PDF Splitter or remove pages you do not need.
Does compressing remove text, links or bookmarks?
No. Text, links, bookmarks and form fields stay as they were. Form fields change only when you tick Flatten form fields, which makes them permanent page content.
Can I compress a signed or password-protected PDF?
Password-protected files are refused because they cannot be changed safely. A digitally signed file can be compressed, but the signature becomes invalid, so keep the signed original.
Sources: PDF 1.7 specification (ISO 32000-1:2008); pdf-lib API documentation.