How to Change PDF Orientation Permanently

You open a PDF and one page is sideways. You rotate it in your reader, scroll on, and forget about it, until you send the file to someone else, who opens it and the page is sideways again. Visual rotation in a reader does not change the file. To rotate a page permanently, you have to save the change back into the PDF itself. This guide walks through the difference between view rotation and content rotation, the tools that do each correctly, and the gotchas that catch people out.

View rotation vs content rotation

A PDF page has a property called /Rotate that tells readers "display this page at this rotation". When you "rotate" a page in Adobe Reader, you might be:

Setting the /Rotate property, affects how every reader displays the page, written into the file.
Rotating in view only, temporary; the file is unchanged.

The first is what you want when you need the change to stick. The second is what some readers do by default for the open-only-this-session case.

Underneath the /Rotate property, there is also the question of content rotation, the actual orientation of the drawing commands on the page. A truly badly-oriented page might have content rotated 90° relative to the page's intended orientation. Fixing one without the other can leave content sideways even when the page reports correct orientation.

The clean fix rotates both: the content stream and the /Rotate property, so the page is genuinely upright in every reader.

Tools that rotate pages permanently

Adobe Acrobat Pro. Tools → Organize Pages → select page → click the rotation icon → save. This sets /Rotate and saves the file. The most reliable GUI option.

Foxit PDF Editor. Page → Rotate. Same workflow.

PDF-XChange Editor. Tools → Document → Rotate Pages.

Preview on macOS. Open the PDF, click a thumbnail in the sidebar, Command+R or Command+L. File → Export to save the rotation.

LibreOffice Draw. Open PDF, right-click page → Rotate. Save as PDF.

Browser-based. Docento.app lets you rotate pages in the browser and save the result without installing anything.

CLI options:

cpdf -rotate-by 90 input.pdf -o output.pdf, rotates every page by 90°.
cpdf -rotate-by 270 input.pdf 5 -o output.pdf, rotates only page 5.
pdftk input.pdf cat 1-3 4east 5-end output rotated.pdf, uses pdftk's syntax to rotate page 4 by 90° clockwise. See pdftk introduction.
qpdf --rotate=+90:5 input.pdf output.pdf, rotates page 5 by +90°. See qpdf introduction.
gs -sDEVICE=pdfwrite -dAutoRotatePages=/None ... for Ghostscript.

For a single page or two, GUI is fastest. For batch jobs across many files, CLI wins.

Rotation values explained

PDFs accept rotation values of 0, 90, 180, and 270 degrees:

0, no rotation
90, 90° clockwise (a portrait page becomes landscape with the top now pointing right)
180, upside down
270, 90° counterclockwise (a portrait page becomes landscape with the top pointing left)

Both 90 and 270 produce landscape orientation but with different "tops". When you have a scanned page that came in sideways, the right rotation depends on which side was scanned first. Try one, see if the text reads left-to-right, and if not, rotate by another 180°.

Common scenarios

A single page came in sideways. Open the file in your editor, find the page, rotate by 90° (or 270° depending on direction), save. Re-share the file.

Every page is upside down. A scan that fed in inverted. Rotate every page by 180°:

cpdf -rotate-by 180 scan.pdf -o fixed.pdf

Some pages are sideways, some are not. Mixed-orientation document, common in scanned books. Use a tool that auto-detects orientation:

ABBYY FineReader has automatic page orientation detection
Tesseract + post-processing can detect page orientation by analyzing text
For high-quality manual fix, use a GUI to inspect and rotate each page

You want to lock the file at a specific orientation. Once rotated and saved, the file reports the new orientation to every reader. Done.

Auto-rotation by content

For scanned PDFs where pages are at random rotations, auto-rotation tools detect which way is "up" by analyzing text:

ocrmypdf --rotate-pages scan.pdf fixed.pdf

OCRmyPDF rotates pages based on OCR's confidence in text orientation. Highly accurate for documents with substantial text. For image-only pages without text, it falls back to leaving them alone.

For more on OCR, see PDF OCR explained and how to make a PDF searchable OCR.

Combining rotation with other operations

Rotation often happens alongside other edits:

Splitting, extract specific pages and rotate them. See how to split a PDF.
Merging, combine documents, then standardize orientation. See how to combine PDF files.
Resizing, change page size first, then orientation, since some tools treat orientation as a side effect of size. See how to change PDF page size.
OCR, rotate, then OCR, since some OCR engines do not handle rotated input well.

A typical scanned-document cleanup pipeline:

Scan to PDF
Run OCRmyPDF with --rotate-pages and --deskew
Verify orientation visually
Manually fix any pages OCR did not rotate correctly
Add bookmarks and metadata
Done

Common gotchas

Rotation looks correct in your reader but wrong in someone else's. Likely you rotated in view, not in the file. Save the file with the rotation baked in and re-share.

Rotation reverts on print. Some printers ignore PDF orientation and reorient based on paper feed direction. Test print to confirm.

Text becomes selectable in the wrong order after rotation. Some PDFs encode text in content-stream order, not visual order. Rotation may make this more obvious. OCR plus tagging fixes it. See tagged PDF vs untagged PDF.

Page numbers and headers/footers misalign. If you added headers/footers as overlays at fixed coordinates, rotating the page may put them in the wrong corner. Add them after rotation, or use a tool that updates header positions during rotation.

Form fields offset. Interactive form fields have absolute coordinates. Rotation can shift them off the visible content. Test forms after rotation; rebuild if needed.

Annotations move. Comments, highlights, and stamps have their own coordinate space. Most tools update annotations during rotation, but verify before committing.

Verifying the rotation took effect

After saving, the surest check:

Close the file
Reopen it in a different viewer than the one you used to rotate (e.g., rotate in Acrobat, verify in Chrome's built-in viewer)
If both show the new orientation, the rotation is in the file
If only the editing tool shows the new orientation, the file did not save correctly, try again

For programmatic verification:

qpdf --show-pages input.pdf | grep -i rotate

This lists the /Rotate value for each page.

What about content that is rotated within an upright page?

A trickier case: the page is portrait, but the content drawn on it is rotated 90°. Reorienting the page does not help, you need to rotate the content stream.

Most tools handle this transparently: when you "rotate page" they rotate both the page property and the content. For tools that only set /Rotate, you may end up with a page reporting one orientation while displaying differently. Acrobat Pro's "Auto-Orient Pages" sometimes catches this; if not, the manual fix is to use a tool that bakes the rotation into the content stream:

cpdf -rotate-contents 90 input.pdf -o output.pdf

(Note the -rotate-contents flag vs -rotate-by.)

Takeaway

Rotating a PDF page is well-supported in every tool, but the key is saving the rotation into the file rather than just rotating the view. For one-off jobs, Acrobat, Foxit, or Docento.app handle it in two clicks. For batch jobs, cpdf and qpdf are reliable. For scanned documents with random rotations, OCRmyPDF can auto-detect and rotate. After saving, verify in a second viewer to make sure the change is persistent, that is the most common point of confusion.