How this page deleter works
Drop a PDF and the grid lays out one tile per page, then fills them with pictures as they draw. Click the circle on a page you want gone and it dims, takes a diagonal line through it, and drops out of the count. Shift-click covers a run of pages in one go. The orange button turns up with your first mark, and it leaves again when you clear the marks or when you cross out the whole document.
Two libraries do the work inside the tab, and both arrive on your first click rather than when the page loads. pdf.js paints the previews. pdf-lib opens the document, detaches each marked page from the page tree, and writes the rest back out. Pages come off from the highest number down, so page 9 is still page 9 when its turn arrives.
Nothing is re-encoded. The text, fonts and images on the pages you keep leave as the same bytes they came in as, and your file never goes anywhere, because the tab you have open is the thing reading it.
When you'd reach for it
- A bank statement with four pages of terms and conditions stuck behind the transactions.
- A scan where the feeder pulled in two blank sheets.
- A contract you are forwarding without the internal cover memo.
- Anything you were about to fix by printing to PDF with a page range, which redraws every page you meant to keep.
What this tool cannot do
Delete every page. Cross out the last one and the button goes away again. pdf-lib saves a pageless PDF without complaint, and loading that file back reports one blank 595 × 842 page where the document used to be. Keep one page and you still have a document.
Make the document much smaller. Cutting 6 of the 12 pages of our test PDF took it from 25,842 bytes to 13,181. Opening the same file and saving it with nothing removed gives 13,199, so the six pages were worth 18 of those bytes. Fonts and images sit in a shared pool the surviving pages still point at; what shrank the file was pdf-lib repacking it, not your deletion. That is why the download row prints one size and no percentage.
Keep bookmarks honest. pdf-lib carries no code that reads an outline tree. Delete page 2 of a document whose contents list points at it, reopen the download, and the bookmark is still there, aimed at an object no page claims any more. Links from one page to another break the same way.
Tidy up a form. A text field whose box sat on a deleted page is still declared in the file afterwards, still holding the widget that used to draw it, and that widget now belongs to a page the document has lost. Some form fillers refuse a file in that state.
Open an encrypted PDF, including the permission-only kind that shows itself in any reader without asking for a password. pdf-lib will not write to a document carrying an /Encrypt dictionary, and this tool cannot take a password off one either. Drop one and the card says so where the grid would have been, so you learn it before you cross anything out.
Leave a digital signature valid. Saving here rewrites the whole document instead of appending to it, and the signature was computed over the original bytes. It survives as an object and fails its check.
Reorder, add or edit. One document per run, and taking pages out is all it does.
Frequently asked questions
Do the pages I keep change?
No. Page size, rotation and content all come through as they were. A page stored sideways is still sideways in the download, and one stored upside down is still upside down.
What does a 300-page file feel like?
The deletion is not the wait. On a 300-page report, five pages come off and the download appears 0.2 seconds after the button. Drawing the grid is the cost: the first preview lands about 0.8 seconds after you drop the file and the three-hundredth about 5.6 seconds in, on a desktop. Pages arrive one at a time, so you can start crossing out the front of the document while the back is still painting.
Can I undo it?
The tool writes a new file and never touches yours, so the original is still sitting in the folder you dragged it from.
Does the document leave my computer?
No. The two PDF libraries download from this site on your first click, and a page holding a JPEG 2000 or JBIG2 scan pulls one more decoder file from here so the preview can draw. Your document itself stays in the tab.
Why is there no button yet?
Either nothing is crossed out or everything is. Taking no page hands you back the same file under a longer name, and taking all of them leaves nothing a reader will open, so the button waits for a mark that sits between the two. A one-page PDF reaches the second state on its first click.
What about a password-protected PDF?
If your reader asks for a password to open it, this tool cannot read it. If it opens without one but still blocks editing, pdf-lib refuses it too and the error says so.