You scan a stack of documents, open the PDF, and half the pages are on their side. You rotate them in your viewer, close the file, send it — and the recipient tells you it is still sideways.
This is one of the most common PDF annoyances, and it comes down to a distinction almost nothing in the interface explains.
View rotation versus stored rotation
Every PDF page carries a rotation attribute stored in the file itself: 0, 90, 180 or 270 degrees. Every viewer honours it when displaying the page.
Separately, most viewers offer a view rotation — a temporary display change that affects only your window. Adobe Reader, Chrome's built-in viewer, Preview's rotate button in some modes, and nearly every browser PDF viewer default to this behaviour.
Rotate the view and the page looks right on your screen. Nothing in the file changes. Send it, and the recipient sees the original orientation.
The tell: if the rotation vanishes when you close and reopen the file, it was a view rotation.
Making it permanent
To change the stored attribute:
The downloaded file has the rotation written into the page objects. Every viewer, on every device, now shows it correctly, because there is nothing left to interpret.
It runs in your browser, so scanned personal documents are not uploaded anywhere.
Why scanners produce this in the first place
Two causes, and knowing which you have prevents a recurrence:
Automatic document feeders pull pages in whatever orientation you loaded them. Load a landscape page into a portrait-oriented feeder and it is captured rotated. The scanner is doing exactly what it was told.
Auto-rotate detection in scanner software tries to work out orientation from the text, and it fails on documents with little text, on forms, on tables, and on anything with mixed orientations. It is right most of the time, which is precisely why the failures are irritating.
If you scan regularly and always get sideways output, turn auto-rotate off in your scanner software and control orientation yourself. Predictably wrong beats unpredictably wrong.
When only some pages are wrong
Common with a mixed stack — a report where the appendix charts are landscape, say. Rotate the affected pages individually rather than everything at once; rotating the whole document just moves the problem to the pages that were fine.
If the pattern is regular — every third page, all the even pages — it usually means the feeder was loaded inconsistently, which is worth fixing at the scanner rather than page by page afterwards.
180 degrees is a different problem
Upside-down pages, rather than sideways ones, almost always mean the paper stack was loaded reversed. Rotating 180 fixes it, but check whether the *page order* is also reversed — if the scanner took the stack from the wrong end, you may need Organize PDF to reverse the sequence as well as the orientation. Fixing the rotation and missing the order produces a document that reads correctly page by page and makes no sense overall.
Rotation and OCR
If you are planning to run OCR on a scanned document, rotate it properly first. OCR engines read horizontally; a sideways page produces either nothing or nonsense. Some engines auto-detect orientation, but not reliably, and it costs nothing to hand them a correctly oriented file.
The same applies to text extraction and to anything automated that will consume the document later.