Four operations that sound identical
Split, extract, delete and reorder all rearrange the pages of a PDF, and the names do very little to distinguish them. Picking the wrong one means doing the job twice — extracting when you meant to split leaves you without the remainder, deleting when you meant to extract leaves you without the part you wanted. This guide gives the distinction in one table, then covers page ranges, what survives the operation, and the cases where none of the four is the right answer.
Which one you want
| Operation | Input | Output | Use it when |
|---|---|---|---|
| Split | One PDF | Several PDFs — every page, or two halves | Breaking a bundle into its parts, or a scan into individual documents |
| Extract pages | One PDF plus a range | One new PDF with only those pages | Sending someone chapter 4 without the other 200 pages |
| Delete pages | One PDF plus a range | The same PDF minus those pages | Removing blank backs, a cover sheet, or an internal appendix |
| Reorder | One PDF | The same pages in a new order | A scan that came out back to front, or an appendix in the wrong place |
| Reverse | One PDF | The same pages, last to first | A duplex scan where the reverse sides came out in the wrong direction |
| Halve pages | One PDF | Each sheet cut into two pages | A book scanned two pages at a time onto one sheet |
Extract and delete are complements: extracting pages 5 to 9 and deleting pages 5 to 9 give you the two halves of the same document. If you need both halves, do it twice on the original rather than trying to get both from one pass.
Page ranges, and where people go wrong
- Page numbers are positions in the file, not what is printed on the page. A report whose body starts at printed page 1 after six pages of front matter has its printed page 1 at position 7. Always count from the first sheet.
- Ranges are inclusive.
3-7means five pages, not four. - Check the total first. Open the document and confirm the page count before typing a range; a range that runs past the end behaves differently in different tools.
- Deleting shifts everything after it. Removing pages 3 to 5 means the old page 6 becomes page 3. Deleting several separate ranges one at a time invites mistakes — specify them all in one go, or work from the end backwards.
- Verify by opening the result, not by trusting the page count. A count of 40 tells you nothing about whether they are the right 40.
What survives a page operation
Page operations copy page objects into a new document, and some document-level structures do not come with them:
| Feature | Survives? | Note |
|---|---|---|
| Page content, text, images | Yes | Copied intact, with no re-encoding and no quality loss |
| Selectable text | Yes | Nothing is rasterised |
| Bookmarks / outline | Often lost | The outline points at pages that may no longer exist |
| Internal links and cross-references | Frequently broken | They reference page objects by position |
| Form fields | Usually kept per page | Fields belonging to removed pages disappear |
| Document metadata | Often reset | The new document is a new document |
| Page numbering stamped on the page | Kept, and now wrong | Re-stamp with Add Page Numbers if it matters |
The last one catches people out: if a document has page numbers printed on it, extracting pages 50 to 60 gives a file whose first page says 50. Add Page Numbers can re-stamp the result, though the old numbers remain underneath.
Common jobs, in order of operations
- Send one section of a long report. Extract the range. Do not split the whole document and fish out the pieces.
- Remove blank reverse sides from a duplex scan. Delete the even pages, or scan single-sided next time. Blank pages are rarely truly blank — scanner noise makes them surprisingly large files.
- Split a bundle of separate documents scanned in one pass. Split into individual pages, then merge the groups back with Merge PDF. Tedious but reliable.
- Fix a scan that is upside down and out of order. Rotate first with Rotate PDF, then reorder. Doing it the other way round makes the thumbnails hard to read while you work.
- Shrink a document that is too big to email. Splitting is a last resort. Compressing usually works better and keeps the document whole.
When a page operation is not the answer
If you want to remove content rather than pages — a paragraph, a name, a figure — no page operation will help, and there is a real risk of assuming otherwise. Deleting the page that shows a name does not remove that name from the rest of the document, and cropping a page usually hides content rather than deleting it. Similarly, if the goal is a smaller file, removing pages saves only what those pages weighed; a single heavy scanned page can outweigh fifty text pages.