Why PDFs Get Big, and How to Shrink Them
How to reduce PDF size: what makes a PDF large, how compression shrinks images and fonts, why some PDFs barely shrink, and how to meet email and upload limits.
By Udit Agarwal · Updated October 18, 2026
To reduce a PDF's size, you shrink whatever takes up the space, and in almost every large PDF that is images: scanned pages or photos stored at a higher resolution or quality than the document needs. Compression tools lower the resolution of those images, re-encode them at a lower quality and drop duplicate or unused data, and the bluntest tools redraw every page as a single picture. Text and vector drawings take very little room, which is why a PDF that is mostly text often cannot get much smaller.
A PDF is a container for page descriptions, fonts, images and extras, and each part weighs very differently:
| Part | Typical weight | Why it grows |
|---|---|---|
| Text and vector drawings | A few kilobytes per page | Rarely the problem: stored as compact drawing instructions and compressed |
| Fonts | Tens of kilobytes to several megabytes | A whole font embedded instead of just the letters used; Chinese, Japanese and Korean fonts are especially large |
| Images | Up to several megabytes each | High resolution, light compression, or photos stored losslessly |
| Scanned pages | Each page is one big image | Scanned at 300 dpi or more, in colour |
| Extras | Varies | Attachments, thumbnails, metadata, and old versions kept by incremental saves |
The last row surprises people. The PDF format allows an edit to be appended to the end of the file instead of rewriting it, so a document edited and saved many times can still carry the objects it replaced until it is saved fresh.
Scans dominate because of simple arithmetic. An A4 page scanned in colour holds this many pixels before any compression:
| Scan resolution | Pixels per A4 page | Uncompressed size |
|---|---|---|
| 300 dpi | 2480 × 3508 (8.7 MP) | 26.1 MB |
| 200 dpi | 1654 × 2339 (3.9 MP) | 11.6 MB |
| 150 dpi | 1240 × 1754 (2.2 MP) | 6.5 MB |
| 100 dpi | 827 × 1169 (0.97 MP) | 2.9 MB |
Compression cuts every row by a large factor, but the ratios between rows hold. Halving the resolution from 300 to 150 dpi leaves a quarter of the pixels, and 72 dpi leaves under 6 percent. Greyscale needs a third of the colour figures, and pure black and white an eighth of greyscale, before compression even starts.
The PDF standard, ISO 32000, lets every image and content stream carry its own compression filter: Flate (the method ZIP uses) for text and drawings, DCT (JPEG) for photos, CCITT and JBIG2 for black-and-white scans, and JPEG 2000. A compressor works on several fronts:
The bluntest method is rasterizing: drawing every page as one image and building a new PDF from those images. It is simple and predictable for scans, but every page becomes a picture. Text can no longer be selected or searched, links and form fields stop working, and crisp vector drawings turn into pixels.
| Method | Lossless? | What you give up |
|---|---|---|
| Removing duplicates and old revisions | Yes | Nothing |
| Subsetting fonts | Yes | Typing new text in that font later |
| Dropping thumbnails, metadata, attachments | Yes, for the pages | The extras themselves |
| Downsampling images | No | Fine detail, visible when you zoom in |
| JPEG recompression | No | Detail, and halos around edges and text at low quality |
| Rasterizing pages | No | Selectable text, links, forms and sharp vector edges |
Lossless steps keep every page exactly as it looks, and they help most with files that were edited many times or embed whole fonts. On scans and photo-heavy files the big savings come only from the lossy steps, so the real question is how much detail you can spare.
A compressor can only remove bytes that are there to remove. A PDF barely shrinks when:
A rasterizing compressor can even make such a file bigger. A page of text that took a few kilobytes as text becomes a full-page image, and the image usually needs more bytes than the text did. If a compressed copy comes out larger, keep the original: it was already small.
Most size problems come from a limit somewhere else. Gmail, for example, allows up to 25 MB of attachments per message and turns anything larger into a Google Drive link. Upload portals set their own limits, commonly somewhere between 100 KB and a few megabytes per file.
Email adds a hidden cost. Attachments travel as Base64 text, which encodes every 3 bytes as 4 characters (RFC 2045), so a file grows by about a third in transit. A 20 MB PDF becomes about 27 MB of message, and a mail server that caps messages at 25 MB will refuse it. The guide to Base64 explains the encoding. Aim well under any stated limit.
When a file cannot get small enough without becoming unreadable, split it. Split PDF by Size cuts a document into parts that each stay under a limit, keeping the pages in order. It does not cut pages in two, so a page that is bigger than the limit on its own needs compressing first.
Quality-based compression asks for settings and shows you the size that results. It suits a file with no hard limit, where you want to judge the balance between size and clarity yourself.
Target-size compression works the other way round: you type the limit, and it searches for settings that fit. Compress PDF to Size offers presets of 100 KB, 200 KB, 500 KB, 1 MB, 2 MB and 5 MB, or any size you type, counting 1 KB as 1,024 bytes. It starts by redrawing every page at 108 pixels per inch and trying JPEG qualities between 24 and 92 percent. If even 24 percent is too big, it lowers the resolution one step at a time, to 81, 61 and finally 46 pixels per inch, searching again at each step until something fits. It makes up to eight full renders and keeps the largest result that fits. If the target is out of reach, it says so and works out the bytes per page the limit leaves. A 100 KB limit across 25 pages leaves 4,096 bytes per page, too little for a readable page image, and splitting the document is the way through.
Compress PDF is a rasterizing compressor. It redraws every page as a JPEG image and rebuilds the document at the original page size. You choose the JPEG quality, from 20 to 95 percent (60 by default), and the resolution: Low draws pages at 54 pixels per inch, Medium at 72 (the default) and High at 108. At Medium an A4 page becomes an image of 595 × 841 pixels, and at High 892 × 1262. The result shows the original size, the new size and the percentage saved, or a warning that the file was already small. Everything runs in your browser, and the PDF is not uploaded.
The limits are the ones rasterizing always brings. Text can no longer be selected, searched or read aloud by a screen reader; links, bookmarks and fillable fields do not carry over; and 72 to 108 pixels per inch reads well on screen but looks soft in print or with small type, so pick High for documents with fine print. It works best on scans and image-heavy files, and it is a poor choice for a text document you need to search. Text in a compressed copy can only come back through OCR (see how OCR works), so keep the original.
PDFs built from photos are good candidates. Image to PDF stores each picture losslessly, which keeps every detail but makes a PDF of phone photos large, and Compress PDF brings it down. Smaller photos make smaller PDFs from the start, so Compress Image to Size is worth a pass first, and iPhone photos saved as HEIC usually need converting before they go in.
How do I reduce a PDF's size without losing quality? Only lossless methods keep every page exactly as it was: removing duplicate images, old revisions and unused data, and subsetting fonts. They help most on files edited many times or with whole fonts embedded. For scans and photos, real savings need lower image resolution or quality, which is lossy.
Why is my compressed PDF bigger than the original? The original was mostly text and vector drawings, which take very little space, and redrawing those pages as images adds bytes. Keep the original; it was already small.
What size should a PDF be for email? Comfortably under your provider's limit. Gmail allows 25 MB of attachments, and Base64 encoding makes a file about a third bigger in transit. For anything near the limit, compress it, split it or share a link.
Does compressing a PDF reduce quality? Lossless steps do not. Downsampling and JPEG recompression do, and how much it shows depends on the settings. At 72 to 108 pixels per inch a page reads well on screen, while small print softens.
Why can't I select the text after compressing? The compressor drew each page as an image, so the text became part of the picture. Keep the original for searching and copying, or recover the text with OCR.
How do I compress a PDF to 200 KB or 1 MB? Use a target-size compressor such as Compress PDF to Size, which searches for the highest quality that fits. If the limit leaves too few bytes per page for a readable page, split the document into parts.
Is it safe to compress a PDF online? It depends on where the work happens. Many services upload the file to their servers. ArrayKit's PDF compressors run in your browser, so the document stays on your device.