How PDF rotation actually works
Unlike a photograph, a PDF page does not need its contents redrawn to appear rotated. Every page carries a /Rotate entry — a number that must be a multiple of 90 — and viewers apply it at display time. The page's text, vectors and images stay exactly where they were.
That makes rotation the cheapest operation in this toolkit. There is no re-encoding step, no quality setting, and no meaningful growth in file size. A 200-page document rotates as fast as a one-page one.
Rotation is cumulative, on purpose
Scanners and phone scanning apps frequently write a rotation entry of their own. When you ask this tool for a 90 degree turn, it adds to whatever was already there rather than replacing it.
The alternative — overwriting the existing value — would mean that rotating a mixed document produces pages that all end up facing the same way regardless of how they started, which is almost never what anyone wants. Adding preserves the relative orientation between pages, so a document where half the pages were already sideways stays internally consistent.
When a rotated PDF still looks wrong elsewhere
Honouring /Rotate is required by the PDF specification, but a handful of older tools and some poorly-written print pipelines ignore it and render the raw page content instead. The symptom is a document that looks correct in your browser and in Acrobat, then comes out of a printer sideways.
If you hit that, the workaround is to flatten the rotation into the content itself by printing to PDF from a viewer that does honour it. You lose the losslessness, but you gain a file that every downstream tool agrees about.
Common uses
- Fixing scanned documents where the feeder pulled pages in sideways
- Turning landscape spreadsheets or tables upright inside a portrait report
- Correcting a phone-scanned document before sending it on
- Preparing a file for a printer that will not rotate for you