Split a PDF into separate pages or pull out the ones you need

The tool asks "How should it be split?" and shows four cards. The question to settle first is whether you want one new PDF or several. One PDF holding only certain pages is Extract pages: type them in "Pages to extract", comma separated with ranges allowed, such as 5 or 12-15, 20. The other three modes always produce a set of files.

Custom ranges is for several files where you decide the boundaries. Each line of the "Ranges, one output file per line" box becomes its own PDF, so 1-4 on one line and 5-last on the next gives two files. A line can also hold a comma-separated list: 2, 7, 9 on one line makes a single file with those three pages, while 2, 7 and 9 on three separate lines makes three one-page files. Fixed size cuts every N pages, N being the number in "Pages per file", with a shorter final file when the count does not divide evenly. Every page needs no settings.

One trap: the tool opens on Custom ranges with 1-1 already in the box, so clicking the button without touching the settings gives you a single file containing only page 1.

Page numbers mean position in the file, counted from 1. They have nothing to do with the numbers printed on the pages. If a report has a cover and a contents page before its printed page 1, that printed page 1 is page 3 to the tool. The line "This document has N pages" that appears once the file is read gives you the count to work from.

The word last (or n) stands for the final page, so 30-last works on any document without counting. Ranges include both ends. Order is kept exactly as typed: 5, 1 puts page 5 before page 1 in the new file, and 10-1 copies that stretch in reverse. A page listed twice appears twice.

Spaces around commas and dashes are ignored, but anything else is not. A semicolon, a slash, or a word other than last or n stops the run, and the message quotes the piece it could not read. A page number past the end stops it too, with a message naming the page count.

The button label counts what your settings would produce, for example "Split into 6 files", and it updates as you type. For Custom ranges it counts non-empty lines, for Fixed size it divides the page count, and for Extract pages it always reads "Split into 1 file". It counts but does not validate: a line with a typo still counts as a file until you click, and the error appears only then. Nothing is written when a line is refused, so fix it and click again.

While it runs, the progress line reads "Building part 3 of 12", one part per output. When it finishes with several files, the panel lists each one with its size and a Download link, plus "Download all as ZIP" for the set, saved as split-pages.zip. Two lines with the same range produce two files with the same name; the ZIP keeps both by adding -2 to the second.

File sizes may not add up. Each new file carries its own copy of the fonts and images its pages use, so one page pulled from a document that shares a large embedded font or a full-page background across every page can still be a large file, and the parts together can total more than the original. Compress PDF can trim the result when the images are JPEGs; it does little for text-only pages.

Document-level items do not travel with the pages. Each output starts as a fresh document, so the original's title and author metadata are gone along with its bookmarks. The pages themselves, including any annotations on them, are unchanged.

Password-protected files stop at the reading step: the tool says the PDF is password protected and points you to Unlock PDF, which needs the password you already have. Above 50 MB the tool warns that the file is processed entirely in memory and may struggle on a phone. Splitting a large scanned document into every page is the slowest case, because each page is written out as its own PDF.