A practical guide

Make a scanned PDF smaller for email

Choose a PDF compressor by privacy, control and limits, using one six-page test as a reference rather than a promise.

Check the limit before choosing a tool

Find the recipient's attachment limit first. Keep the original PDF, and aim a little below the limit because email systems can add overhead. A smaller file is useful only if the recipient can still read it.

Scanned pages are mostly images, so reducing image resolution or quality can make a large difference. A text-only PDF may shrink much less. If the file contains private or regulated material, check your organization's rules before uploading it to any online service.

What happened to the same test file

On 5 October 2026, we made a six-page, entirely synthetic scan-style PDF with small print. It was 5,317,166 bytes (5.32 MB, decimal). We uploaded this same file to each service without signing in or paying and downloaded the results. All three outputs retained six A4 pages; the small print on page one remained readable in a 1,200-pixel full-page render.

One sample and one run per service. Output size is the downloaded file, not a pre-run estimate.
Tool and settingOutputReductionWhat to know
PDF24 — 144 DPI, 75% image quality626,429 bytes (0.63 MB)88.22%Default controls in this run; you can adjust DPI and image quality.
iLovePDF — Recommended496,536 bytes (0.50 MB)90.66%Three visible compression levels; this run used the selected Recommended level.
Smallpdf — Basic441,780 bytes (0.44 MB)91.69%Basic worked without an account in this run; Moderate and Strong are marked Pro.

Smallpdf's Basic output was smallest on this one invented file. That is a result for this sample, not a general ranking: images, fonts, forms and previous optimization can change both file size and visual quality.

Choose by the constraint that matters to you

If the document cannot leave your device, use an approved offline workflow. PDF24 offers a separate Windows Creator app according to its documentation, but we did not install or test it. None of the three online tests establishes that uploading a confidential file is permitted or safe for your situation.

For an ordinary one-off file that you may upload, Smallpdf Basic is a simple free starting point and produced the smallest file in this test. Its site marks Moderate and Strong as Pro and says free usage limits may apply.

Choose iLovePDF if you prefer three named compression levels. Recommended is a sensible first try when you want an easy quality-versus-size choice.

Choose PDF24's online tool if you want numeric DPI and image-quality controls instead of named presets. Its default output was larger on our sample; the extra control is the reason to choose it here.

If the first output is still above the recipient's limit, change one available compression setting at a time and check legibility again. If the smaller file becomes hard to read, keep the readable version and ask the recipient for an approved file-sharing route instead of forcing the file below the limit.

  • Smallpdf — Try Basic for a simple, free one-off run; check the current free-use limits before repeat work.
  • iLovePDF — Try Recommended first if you prefer named compression levels; inspect the output before choosing Extreme or Less compression.
  • PDF24 Tools — Choose the online tool when you need direct DPI and quality controls. Its separate Windows Creator is an offline option described by the vendor, not tested here.

Check the downloaded PDF before sending

Open the downloaded file and check its actual size against the recipient's limit. Inspect the smallest text, images and every page at a comfortable reading zoom. If the PDF includes forms, annotations, links, signatures or attachments, verify those parts still work; our sample did not contain them.

All three online tools discussed here process uploaded files on their servers according to their own documentation. For sensitive documents, an approved offline workflow is the safer starting point. PDF24 says its Windows Creator works offline; we did not install or test that app.

What this test does and does not prove

This was one synthetic, image-based PDF on one day, with one run per tested setting. We checked downloaded byte size, page count and the first page's visible small print. We did not benchmark speed, test text-only or protected PDFs, check every page's visual detail, or independently audit the services' security and deletion claims. Product plans and limits can change; the linked official pages were checked on 5 October 2026.