To compress a PDF to a specific file size, you need a target-aware tool that verifies the actual bytes of the saved output before it offers a download, not one that guesses from the source file size. Compress PDF to Size is built for exactly this: you enter a target between 0.1 MB and 25 MB that is smaller than your source PDF, and the tool runs a short, bounded sequence of JPEG quality and maximum-dimension profiles against eligible embedded JPG streams. After each profile it saves the candidate PDF, measures its real byte length, and compares that number to the target you typed — no estimates, no rounded approximations. A download link only appears when a candidate actually meets or beats your target; if no safe profile can get there, the tool creates no download and tells you why. Because every read, re-encode, save, and verification step stays inside your browser, the source file never leaves your device.
How Target-Size Compression Works
The difference between a target-size tool and a generic PDF compressor comes down to what the tool actually promises to check. A generic compressor runs one fixed profile, hands you a smaller file, and labels it "compressed" — but it has no concept of your target. A target-aware tool compares the saved bytes to the number you entered before it offers a download.
Compress PDF to Size follows the second pattern. You enter a target MB value between 0.1 and 25 that is smaller than your source PDF, and the tool iterates through a fixed sequence of four profiles. Each profile pairs a JPEG quality setting with a maximum image dimension, and the tool only swaps in a re-encoded JPG stream when the replacement is actually smaller than the original. After every profile, the candidate PDF is saved, its byte length is measured, and that number is compared to your target.
| Profile | JPEG Quality | Longest Image Side |
|---|---|---|
| 1 | 0.70 | 1400 px |
| 2 | 0.55 | 1000 px |
| 3 | 0.40 | 800 px |
| 4 | 0.30 | 600 px |
The first candidate that lands at or below the target is the one you can download. If none of the four profiles can hit the target safely, you get no download and a plain reason why. Because the comparison is done on the real saved file and not on an estimate derived from the source, the MB value in your file properties is the MB the tool actually matched. That is the core guarantee, and it is what makes the tool useful when you need to hit an email attachment cap, a portal upload limit, or an application form's maximum size rather than simply make a file smaller.
Common Size Limits You'll Run Into
Most size caps fall into a few predictable buckets, and knowing yours up front helps you set a realistic target before you start. A typical mapping between target size and where the limit comes from looks like this:
| Target Size | Typical Reason | Source Type That Reaches It |
|---|---|---|
| 1 MB | Many application and government portals | Scanned or photo-dominated PDFs |
| 2–5 MB | Job application systems, email gateways | Image-heavy PDFs with eligible JPG streams |
| 10 MB | Common webmail attachment caps | Most JPG-heavy source PDFs |
| 20–25 MB | E-signature services and large-file portals | Most PDFs with eligible JPG content |
Beyond the upload-cap range, the 25 MB source ceiling on Compress PDF to Size covers nearly every common 0.1–25 MB target. For larger source files that still need to land under a portal limit, Split PDF by Size divides one local PDF into parts measured against a maximum serialized file size, and each part can then be compressed separately to fit the same portal cap.
Compress a PDF to a Target Size
Here is the exact sequence the tool is designed for.
- Open the Compress PDF to Size tool in your browser.
- Choose one local PDF file. The source limit is 25 MB, and only one file is processed at a time.
- In the target field, enter a size in MB that is smaller than your source file. The accepted range is 0.1 MB to 25 MB.
- Click "Compress to target size." The tool loads the file and begins the bounded profile sequence.
- Wait for the comparison to finish. If a profile's saved output is at or below your target, a download link appears.
- Download the file only after the link is shown. If no candidate meets the target, the tool explains why and offers no download.
Each step is a real operation against the saved file. The PDF is read in your browser, eligible JPG streams are re-encoded through the browser's canvas pipeline, the candidate is saved, and the comparison reads back the actual file length before deciding whether to expose the result. Because the source is never uploaded, you can disconnect from the internet after step 2 and the rest of the run still completes normally.
What the Tool Changes Inside the PDF
To meet a target safely, Compress PDF to Size restricts itself to a narrow category of edits. It only re-encodes embedded JPG image streams that meet three conditions at once:
- The stream uses DCTDecode, the standard JPEG decode filter.
- The image's color space is DeviceRGB or DeviceGray.
- The stream has no decode parameters and is not masked.
The local job is also bounded by an 80-image ceiling and a 100-megapixel total across all eligible images, so very large photo-heavy PDFs are processed within a fixed budget rather than unlimited work. If an image does not match the three eligibility conditions, the tool leaves it exactly as it was. If a re-encoded JPG would actually be larger than the original stream, the tool also leaves that image alone — it never makes a file bigger in the name of "more compression."
Everything that is not an eligible JPG stream is preserved. Page structure, selectable text, links, form fields, vector content, and unsupported image objects (CMYK JPEGs, JPEG2000/JPX, JBIG2, and any masked image) remain unchanged. The tool does not rasterize whole pages, which means your text stays as text and your vectors stay sharp.
The kind of PDF you start with therefore has a direct effect on how low a target you can hit. A scanned document, a photo album exported to PDF, or a slide deck built from screenshots is dominated by RGB JPGs and tends to compress well. A text-only report, a vector-heavy brochure, or a PDF built from CMYK print assets may not move much at all, because most of its bytes are not in eligible JPG streams.
When a Target Cannot Be Reached Safely
There is a hard ceiling on what bounded JPG recompression can deliver, and the tool will tell you when you hit it. If none of the four profiles produces a candidate at or below your target, no download is created. The reasons usually fall into three buckets:
- The PDF is text-only or vector-heavy. With little or no eligible JPG data to re-encode, the four profiles cannot meaningfully shrink the file.
- The dominant image format is unsupported. CMYK JPEGs, JPEG2000, JBIG2, or masked images are not eligible for replacement, so the tool has no JPG payload to work on.
- Your target is too aggressive for the content. Asking for 0.5 MB from a 20 MB brochure of CMYK photography will exceed what safe JPG re-encoding can deliver, even at the lowest profile.
When this happens, your practical options are to relax the target to a larger number, use Split PDF by Size to break the file into parts that each fit the limit, or convert the source to a JPG-first PDF using JPG to PDF so the images are eligible from the start. The tool itself does not invent a smaller file or claim approximate success — it stops and explains.
Checking the Result Before You Share It
Once a download link appears, the saved file is what you should measure and review, not what the tool claims.
For a quick check, right-click the downloaded PDF, choose Properties on Windows or Get Info on Mac, and confirm the file size against the target you entered. Because the tool only exposes the download after measuring real bytes, the number on disk will match or beat your target — but checking confirms what is actually saved.
For a deeper review, open the file in a viewer and confirm the following:
- The page count matches the original. The tool reopens each candidate with a PDF reader and checks the page count before exposing the download, and a final visual check confirms nothing was dropped or duplicated.
- Selectable text still selects. If your original had a text layer, it should still be there.
- Links and form fields still work. Open a representative link and a representative form field.
- Embedded images look acceptable. Because re-encoding happens at the JPEG layer, expect some quality reduction on image-heavy pages — this is the trade that lets the file hit your target.
If anything looks wrong, your original file is still on disk — the tool does not modify it — so you can re-run with a higher target or try a different approach.