How to remove pages from a PDF

How to delete pages from a PDF

Blank backs from duplex scanning, a duplicated sheet, a draft version of an annex, a section this particular recipient has no business reading. The operation looks trivial, and nine times out of ten it is. The tenth case is either the wrong sheet gone or a table of contents that fell apart, and both are predictable.

Two strategies, and the second is often shorter

The tool has a strategy switch, and it inverts the meaning of every tick you make on the preview grid.

The first is the default. The second is not a different tool, it is the same one with the ticks inverted, and the choice is pure arithmetic: ticking six pages beats ticking thirty-four.

A limitation hides here that is easy to miss. In keep-selected mode the surviving pages are forced back into the document's original order, no matter what order you ticked them in. Ticking 7, then 2, then 5 will not produce that sequence; the output holds 2, 5, 7. The tool can only remove, and reordering belongs to organize-pdf.

How to drop the sheets

1. Open remove-pages and upload the document. 2. Wait for the preview grid; it is there so you tick by picture rather than by number. 3. Switch the strategy if the shorter list is the one you want to keep. 4. Tick the pages, or type a range if that is faster. 5. Run the job and download the result. 6. Open the file and check the seams: the page before and the page after every gap you created.

Checking seams beats paging through the whole file. A selection mistake almost always shows up exactly at a boundary: a sentence breaks off, a table starts halfway down, a cross-reference points at nothing.

How a page selection is written

Besides ticking with the mouse, the selection can be typed, and the format is strict: items separated by commas, each item either a number or an interval joined by a hyphen. Writing `2,5,9-12` names six pages: the second, the fifth, and four in a row from nine to twelve. Numbering starts at one, and repeats are dropped, so `3,3,3` means the same as `3`.

Strictness matters more than convenience here, and deviations are not forgiven quietly:

That last row is the common one in practice. Numbers written down for the original document stop being true the moment you process it once: the file got shorter and the note did not. The job will not run partially and will not ignore the surplus, it stops outright. That beats silent truncation, but it does mean recounting the list after every edit.

A guard against deleting half the document

The parameters hold a switch that demands a preview, off by default. Turned on, it blocks the job whenever more than half the pages would go, and says so with its own error rather than processing quietly.

Its purpose is narrow and concrete. A mistake in a bulk selection does not look like a mistake: the file arrives, it opens, the pages are there, there are simply four times fewer of them than there should be. For signed contracts, powers of attorney and court annexes, that check is worth one extra click.

The other side of it: if you deliberately keep six pages out of forty, the guard fires at you too. Then switch it off rather than hunting for a way around it.

One limit cannot be switched off at all: you cannot remove every page, and you cannot process a single-page document. An empty PDF is formally possible and practically useless, and the tool will not hand you one.

What happens to the outline and the document properties

The output is not a bare stack of pages. The file's properties (title, author, keywords) carry over from the source unchanged. Worth remembering when pages are being removed for confidentiality: the author's name and the original title stay in the properties, and clearing them is a separate job for set-pdf-metadata.

The outline carries over with its targets recomputed. Every bookmark points at its page's new position. If the page is gone, the corresponding entry is dropped while its children move one level up and stay in the contents. The reasoning is simple: deleting one sheet that happened to carry a section heading should not delete the section from the table of contents.

The practical consequence is worth a minute: after removal, open the bookmarks panel and walk it. The structure survives, but the hierarchy can come out one level flatter wherever section title pages were the ones removed.

When the job belongs to a different tool

Several requests that look like page deletion resolve faster another way.

Dozens of blank backs after duplex scanning are not worth ticking by hand; remove-blank-pages exists for that, with an adjustable sensitivity and a blankness threshold.

If you need a handful of pages and do not care what happens to the rest, extract-pages reaches the same result in fewer moves and can also emit one file per page.

If the document has to become several standalone files rather than a cleaner single one, that is split-pdf.

And last: a document locked with an open password cannot be processed at all; the job stops with a direct message asking for the password. Clear the protection with unlock-pdf and come back. A password that only restricts printing or copying does not get in the way.