Delete PDF pages

Delete and Extract are mirror images of each other. Delete keeps everything you do not select and discards what you do select. Extract keeps only what you select and discards everything else. If you want pages 1, 2, 4, 5 out of a 5-page PDF, you can mark page 3 in Delete or mark pages 1, 2, 4, 5 in Extract — both produce the same result. Choose Delete when you are removing a few pages from a long document (the deletion list is shorter). Choose Extract when you are pulling a small slice out of a long document (the extraction list is shorter). The output files are identical in either case, since both tools use native page-copy with the same options.

Yes. Marking is reversible — click a marked thumbnail again to unmark it, or use the Clear marks button to reset everything. Nothing is actually deleted until you press Save, so you can experiment freely. The original PDF stays loaded in browser memory; the delete operation produces a new file rather than modifying the upload. If you save and realize you removed the wrong page, the original is gone from the saved output, but as long as you have the source file on disk, you can reload it and try again. Mistakes are cheap because nothing is permanent until you both save and overwrite the original on your filesystem, which the browser cannot do automatically.

Yes. Pages you keep retain all their original content: text, images, annotations, form fields, hyperlinks, and embedded fonts. Internal links that pointed at deleted pages become dead (the link still exists but its destination no longer exists in the document), so it is a good idea to delete pages before applying signatures or finalizing a deliverable. Bookmarks (outline entries) referencing deleted pages are removed from the outline tree automatically. Annotations and form fields attached to deleted pages are gone with them — there is no way to keep an annotation that lived only on the removed page. The remaining form is still functional, but if the deleted pages were part of a multi-page form, you may need to renumber field labels manually.

The most common cause is forgetting to press Save after marking pages — marks alone do not modify the file. Check that the count on the Save button matches the expected remaining-page total (not the deleted-page total). Another cause is loading the wrong file back into a viewer: browsers cache files by name, so if you saved over an existing file and reopened it, your viewer might still be showing the cached old version. Refresh the viewer or rename the output before opening. If neither applies and the page count truly does not change, the PDF may contain layered or alternate page objects (uncommon, but it happens with old scanner output) — try the Organize PDF tool, which uses a slightly different page-extraction path.

Encrypted PDFs need to be unlocked first using the Remove PDF password tool with the correct password. Once decrypted, the saved copy can be loaded into Delete PDF pages normally. Signed PDFs can be loaded and modified, but any signature on the original will show as invalid in viewers after pages are removed, because the signature covers the entire byte content of the file. If preserving signatures matters, do not delete pages after signing — re-sign the trimmed document instead. The tool will warn you when it detects a signature, but it will not block the operation. PDF/A documents can also be modified, but the result may lose its PDF/A conformance flag and need re-validation if you are submitting to an archive.

No fixed maximum is set by the tool, since the work happens in browser memory. Practical limits depend on your device: a recent laptop with 8 GB or more of RAM handles PDFs up to several hundred megabytes without trouble. Scanned PDFs with high-resolution images consume far more memory than text PDFs at the same page count, so a 500-page text PDF may be lighter than a 50-page scan. If the file is so large that thumbnails fail to render, use the text-input mode to type page ranges (like 1, 5, 10-12) without scrolling thumbnails — the delete operation itself uses much less memory than the preview grid. None of the data is uploaded; the only cost is your device's RAM during the session.