How to split a PDF into pages or page ranges

How to split a PDF into pages or ranges

A counterparty needs three pages out of a thirty-page contract. An annual report has to yield one section. A heavy scan will not clear the intake form and has to come apart into lighter pieces. Different jobs, and each has its own split mode.

One rule settles half the confusion: if the output has to be a single file, this is not the tool. If you simply need pages 4, 9 and 15 in a single document, that is work for extract-pages. If you need to drop a couple of sheets and leave everything else intact, remove-pages handles it. Splitting is for when the result genuinely has to be several standalone documents.

Six modes and the jobs behind them

Pick the mode from the "Split mode" list. Only the fields that mode needs appear afterwards: you will not be asked for ranges unless you chose the ranges mode.

Step by step

1. Open split-pdf and upload the document. 2. Wait for the thumbnails and use them to find the real section boundaries. 3. Choose the mode in the "Split mode" list. 4. Fill in the field that appears: ranges, a page count, or a size ceiling. 5. Run the job and download the archive. 6. Unpack it and open the outer pages of each part. The middle is usually fine; mistakes live at the boundaries.

How ranges are written

The format is simple: intervals separated by commas, pages within an interval joined by a hyphen. Writing `1-3,8,10-12` yields three files: pages one through three, a single page eight, and pages ten through twelve.

Numbering offset

The most common mistake when splitting is not the syntax but which number you mean. The number printed on the page and the page's position inside the file rarely match: a title sheet, a reverse side, a contents page each shift the count by a couple of positions. In a contract where the sheet reads "5", the PDF may hold it as page seven.

So find boundaries by thumbnail, not by printed numbering. For a document up to eighty pages every thumbnail renders at once. Beyond that, the far ones draw as you scroll, so you have to reach the spot and wait a second or two.

What overrides the chosen mode

Here is a rule that is easy to miss: ranges marked directly in the page editor take priority over the mode picked in the list. If you first selected pages by hand and then switched the mode to "every N", the manual selection wins.

That is convenient when you have already marked the document up and do not want to lose the markup. It is also confusing when you switched modes deliberately and the result came back unchanged. If the behaviour does not match what you set, clear the manual page selection first.

Splitting by weight rather than by meaning

The "by file size" mode solves a different problem: it does not matter where the meaningful boundaries fall, only that each part fits the limit. The threshold is given in kilobytes and defaults to 1024, that is one megabyte.

Next to it sits a switch that permits compression. With compression the parts hold more pages, because each one is squeezed down to the required weight. Without it the document is cut more often but the content stays untouched. For scans people will read, prefer the first. For documents heading into an authenticity check, the second is safer.

Even and odd, for double-sided scans

This mode exists for one specific misery. A sheet feeder without duplex runs the whole stack face up, then someone flips it and runs the backs. The result is a file with all the fronts first and all the reverses after, sometimes in reverse order.

Splitting into even and odd gives exactly two files: one with pages 1, 3, 5 and onward, the other with 2, 4, 6. You then interleave them correctly with merge-pdf. The count here is always two files, however many pages the source held.

The output is an archive

The result arrives as a single ZIP with the parts inside. That applies to every mode: even the even and odd split, which produces just two files, comes packed in an archive.

Part names depend on the mode and nearly all of them carry the source page numbers. That is handier than it sounds, because the name tells you what is inside without opening it.

Numbers everywhere carry leading zeros. That is not decoration: without them part ten would sort ahead of part two, and reassembly would mean restoring the order by hand.

In the size mode a name sometimes ends in `_compressed`. That marks the part that had to be squeezed to meet the weight you set, while the others stayed untouched. If quality turns out to have suffered somewhere, look in the files carrying that suffix.

If you want a single document rather than an archive, switch on "Merge selected ranges". The intervals you named then come back as one PDF instead of a ZIP. The option is built for the modes where you set ranges or mark pages by hand, and means nothing for a split into single pages.

How many parts you will get

Splitting caps the number of parts, and not out of spite: a thousand-page document in single-page mode would mean a thousand files to create, compress and pack.

If the cap is exceeded the job does not start, and the message tells you both how many parts you would have got and what the maximum is. No guessing needed: take a coarser mode. Every N pages replaces single pages; wider ranges replace narrow ones.

The check fires before a single file has been created, so a doomed job drops out immediately instead of burning time on parts that would be thrown away. In the size mode the count also runs during packing, since the number of parts is not known in advance.

When you do not need a split at all

Sometimes the job looks like splitting but resolves faster another way.

Separate pages in one file: that is extract-pages. Dropping the surplus while keeping the rest whole: remove-pages. Swapping sections around inside a document: organize-pdf. Reach for splitting when the output genuinely has to be several separate files rather than one modified version of the original.