Splitting a PDF sounds simple, but people get stuck in three places: how to write the page ranges without an error, why the parts add up to more than the original file, and where to cut when a file is too big to send. This article works through all three with ToolGarden's Split PDF tool. Everything runs locally in your browser, and your file is never uploaded.
First, decide which kind of split you need
"Split" covers three different needs, each with a different approach. Pick the right one and it is a single step; pick the wrong one and you redo it.
| Scenario | Approach | In the tool | Output |
|---|---|---|---|
| Send chapters to different people | Split by page ranges | Type 1-5; 6-18; 19-30 and click Split by ranges | ZIP with 3 PDFs |
| One file per page | Every-page export | Click Split every page; no input needed | ZIP, one PDF per page |
| Keep a few pages, drop the rest | Extract pages | Use Extract Pages and type 2,7,15-17 | One new PDF |
| File too big to send or upload | Split by size (indirect) | Estimate pages per part, then split by ranges | ZIP of smaller parts |
| Paste pages into chat or docs | Export images, not split | Use PDF to Image for PNG or JPG | Images, optional ZIP |
Note the last two rows. The Split PDF tool has no "split by size" option; it accepts page ranges only. Splitting for a size limit means planning the page numbers yourself. A full section below covers how, and where that approach stops working.
Steps: running one split with the Split PDF tool
- Open Split PDF and click the upload area or drop a PDF onto it. The tool reads the file immediately and shows its size and page count under the file name, for example "4.2 MB · 48 pages".
- The range box is pre-filled with an example (for documents over one page, 1-2; 3-last). Replace it with your own plan.
- Click Split by ranges. For one file per page, click Split every page instead.
- The result area lists each output file with its name, page range, and page count, plus the total size and time taken. Check it, then click Download to get a ZIP.
If the input is invalid, such as a page number above the total, the tool shows a red message under the range box saying the page is outside 1-48, and produces nothing. (These range-validation messages are currently written in Chinese, whatever the site language.)
Range syntax: four rules
The tool thinks in groups. Each group becomes one separate PDF. Groups are separated by a semicolon or a line break, and inside a group you list pages with commas and hyphens.
1-3; 4-6; 7 three files: pages 1-3, pages 4-6, page 7
1-3
4-6
7 same thing, using line breaks instead of semicolons
1,3,5-7; 2,4 two files: pages 1,3,5,6,7 and pages 2,4
10-12; 1-9 first file is pages 10-12, second is 1-9 (file order follows input order)- Rule 1: pages are 1-based physical positions in the file, not the number printed on the page. For books with roman-numeral front matter, use the "page N of total" your reader shows.
- Rule 2: ranges must go low to high. 5-3 is rejected as an invalid range.
- Rule 3: within one group, a repeated page appears once and keeps the order you typed. Groups are independent, so the same page can appear in several groups, which copies it into several files.
- Rule 4: full-width commas and semicolons used in Chinese text are normalized, so a pasted "1-3;4-6" still works.
Example 1: a 48-page annual report for three teams
Finance wants pages 1 to 5 (summary) plus pages 30 to 36 (statements). Marketing wants pages 6 to 18. Management wants just the first page and the last two, 47 and 48. Finance needs non-contiguous pages and management wants head and tail, which is exactly what commas inside a group and semicolons between groups are for.
Range box:
1-5,30-36; 6-18; 1,47-48
Output (file name pattern: original-part-N-pagelabel.pdf):
report-part-1-1-5,_30-36.pdf -> 12 pages (5 + 7)
report-part-2-6-18.pdf -> 13 pages
report-part-3-1,_47-48.pdf -> 3 pagesThe page label in the file name is built as "1-5, 30-36", and any character other than digits, commas, and hyphens (the space) is replaced with an underscore, so an underscore appears after each comma. That is expected. The result area shows the page count for each part, so you can check it against your plan: if 12, 13, and 3 do not match, a page number was mistyped.
Example 2: one file per page for batch archiving
Take a 30-page scan where each page is a separate invoice. Click Split every page and the tool produces 30 files named scan-part-1-1.pdf through scan-part-30-30.pdf, bundled in scan-split.zip. Afterwards you can rename them by invoice number with any bulk renamer. One detail: because the sequence number comes before the page label, a file manager sorting by name will place part-10 before part-2, since it is a string sort. Pad the numbers during bulk renaming, or sort by modified time.
Example 3: splitting a manual with front matter by chapter
Take a 200-page product manual with a 1-page cover and 3 pages of contents, then body text starting at physical page 5, with printed numbering restarting at 1. Chapter 1 is printed pages 1 to 40, chapter 2 is 41 to 95, chapter 3 is 96 to 190, and the appendix is 191 to 196. To split into the front matter plus four chapters, first convert printed numbers to physical positions: physical position = printed number + 4.
Front matter (cover+contents) physical 1-4
Chapter 1 printed 1-40 physical 5-44
Chapter 2 printed 41-95 physical 45-99
Chapter 3 printed 96-190 physical 100-194
Appendix printed 191-196 physical 195-200
Range box: 1-4; 5-44; 45-99; 100-194; 195-200
Expected output pages: 4, 40, 55, 95, 6, total 200A total of 200 is the best self-check for this kind of plan: if the ranges are contiguous, non-overlapping, and end to end, their page counts must sum to the document total. If the sum is off, one range is wrong, and you find out before clicking split.
Splitting by size: the 25 MB email limit
Many mailboxes and upload forms cap attachments, and 25 MB is a common figure (check your own service). The splitter has no "each part under X MB" option, but estimate-then-verify gets you there.
Step 1: estimate average page weight
Divide total size by page count. Say a 120-page scanned contract is 60 MB, so 0.5 MB per page. To stay under 25 MB, a part could hold about 50 pages, and 40 leaves headroom. The ranges become 1-40; 41-80; 81-120.
Step 2: verify and adjust
Averages lie. In one document, image pages and text-only pages can differ in weight by a large factor. After splitting, look at each file's size in the result and re-split any part that is over. The table below is an illustration (the numbers are made up to show the method):
| Output file | Pages | Expected | Actual (example) | Action |
|---|---|---|---|---|
| part-1 | 1-40 | ~20 MB | 14 MB | OK |
| part-2 | 41-80 | ~20 MB | 31 MB | Over; split into 41-60 and 61-80 |
| part-3 | 81-120 | ~20 MB | 15 MB | OK |
One counter-intuitive fact: the parts usually add up to more than the original. Fonts, images, and other resources shared by many pages are stored once in the original, but after splitting each output file must carry its own copy of whatever it uses. The more pieces, the larger the total. If your goal is simply a smaller file rather than distribution in pieces, splitting is not the best tool; reduce image resolution or rebuild the pages through image export instead.
Five checks after splitting
- Sum of pages: the page counts of all outputs should equal the number of pages you specified (the original page count, if no page repeats).
- First and last pages: open two files and confirm the first and last pages are what you intended, which catches an off-by-one in the ranges.
- File names: confirm the numbering and page labels are clear, and rename them to something the recipient will understand before sending.
- Size: for size-limited uploads, confirm the largest file is within the limit.
- Internal links and bookmarks: table-of-contents jumps in the original may break, because their target pages may now live in a different file.
Common problems and diagnosis
| Symptom | Cause | Fix |
|---|---|---|
| Message: PDF could not be read; it may be damaged or encrypted | The file has an open password or is damaged | Decrypt it first in the PDF Encrypt / Decrypt tool, then split |
| Message: page N is outside 1-M | Page number above the total | Check the page count shown under the file name and retype |
| Message: page range 5-3 is invalid | Range written backwards | Write 3-5 |
| Message: file too large | A single file is over 80 MB | Compress it elsewhere first, or work in sections |
| Parts have no bookmarks | The tool copies pages, not outlines | Add bookmarks in desktop software if needed |
| Form fields missing in the parts | Form field definitions are document-level and are not copied with pages | Flatten the form to PDF first, or re-add fields after splitting |
When splitting is the wrong tool
- You only need a few pages: Extract Pages is more direct and gives one file instead of a ZIP.
- You hope to reduce file size: see above; splitting often makes the total bigger.
- The file carries a digital signature: splitting produces a new file and the original signature stops being valid.
- You want automatic splitting by chapter title or bookmark: the tool takes page ranges only and does not read bookmarks. Read your table of contents and write the ranges yourself.
- Very large page counts or a file near 80 MB: browser memory gets tight, so process in several passes.
Privacy
The whole process happens in your browser: the file is read into memory, split, and the ZIP is built locally for you to download. Watch the developer tools Network panel and you will see no request carrying file contents during processing. Sensitive files such as contracts, medical records, and financial data can be split without leaving your machine.