Rotate PDF Pages

Rotate one or all pages in your PDF file. Pick 90, 180 or 270 degrees. Free, browser-based tool.

✓ Free✓ No sign-up✓ Works in browser

Drop a PDF here or click to browse

Adobe Acrobat

The world standard for PDFs

Partner

Edit, sign, compress, merge, protect PDFs. 500M+ users. 7-day free trial.

Start Free Trial

How to Use This Tool

Upload Your PDF

Upload the PDF with pages that need to be rotated. Page thumbnails are displayed for all pages.

Select Pages and Rotation

Choose specific pages or 'All pages'. Select the rotation angle: 90° clockwise, 90° counter-clockwise, or 180°.

Save and Download

Click Apply and download your PDF with the corrected page orientation.

Related Tools

Frequently Asked Questions

Can I rotate only some pages, not all?

Yes. Select individual pages or a page range to rotate. Other pages remain unchanged.

Does rotating affect image or text quality?

No. Rotation is a metadata operation that changes the page orientation without re-rendering or re-compressing any content.

Why are my PDF pages sideways?

This commonly happens when scanning documents on a flatbed scanner or mobile camera at the wrong orientation. Rotating the PDF pages corrects this.

Is there a page limit?

No. The tool handles PDFs of any length.

About Rotate PDF Pages

Your document scanner fed three pages sideways into a 40-page contract, and the recipient keeps emailing back 'can you fix the orientation on pages 12, 17, and 29 so I can actually read them'. Or you exported a spreadsheet to PDF and the last page came through in landscape while the rest is portrait, making the combined file awkward to print duplex. This rotator edits the /Rotate attribute on individual page dictionaries via pdf-lib — it does not rasterize pages or re-render text, so vector fonts stay sharp, file size barely changes, and OCR text layers remain intact. You can rotate 90, 180, or 270 degrees per page, apply to selected ranges or all pages, and the output preserves bookmarks, form fields, and digital annotations. Runs entirely in the browser tab with no server round trip, which matters when the document contains account numbers or PHI that cannot leave your device. The operation is metadata-level, so a 200-page rotation completes in well under a second even on a mid-range laptop.

When to use this tool

Fixing a scanner feed error

An office worker scans a 40-page contract through an automatic document feeder that grabbed three pages sideways. Identify pages 12, 17, and 29 from the preview, apply 90-degree clockwise rotation to each, save the corrected output, and re-send to the counterparty without re-scanning the entire stack.

Normalizing mixed-orientation exports

A finance analyst exports a multi-sheet Excel workbook to PDF. The income statement pages are portrait but the wide cash-flow schedule came out landscape. Rotate the landscape pages 90 degrees so the entire 20-page packet prints consistently in portrait for the board binder.

Correcting a phone-camera document scan

A field tech photographs a service report on an iPhone, the PDF export rotated the image 90 degrees based on wrong EXIF metadata. Rotate the single page, save as a fixed 1-page PDF, email to dispatch before the tech moves to the next job site.

Preparing pages for duplex printing

A print-shop operator has a PDF where odd pages are portrait and even pages are rotated 180 degrees (so duplex printing produces right-side-up output on both sides). Batch-rotate the even pages 180 degrees so the entire file prints normally on a standard duplex-capable printer.

Restoring orientation after a bad OCR

An archivist runs Tesseract OCR on a historical document scan; the OCR engine auto-rotates some pages incorrectly (it interpreted a logo as the top of the page). Identify the misrotated pages, rotate back 90 or 180 degrees, re-save. The underlying text layer persists because rotation is metadata-only.

How it works

1
Rotation is a metadata flip, not a render
PDF stores page orientation as the /Rotate attribute on each page dictionary — a value of 0, 90, 180, or 270 that the viewer applies at display time. pdf-lib.js reads the current /Rotate value (defaulting to 0 if absent), adds your requested delta modulo 360, and writes back the new value. No pixel re-rendering, no canvas round-trip, no font re-embedding.
2
Inherited rotation from page tree is handled
PDF allows /Rotate to be inherited from ancestor nodes in the page tree (a /Pages node can set a default that all children inherit unless they override). pdf-lib walks the inheritance chain to compute the effective current rotation, applies your delta to the effective value, and sets the new rotation on the specific page rather than the parent — so rotating one page does not silently rotate siblings that were inheriting the same default.
3
Output is re-serialized with the same xref format
After rotation the document is re-serialized through pdf-lib's save() which preserves the original cross-reference format (classic xref table for PDF 1.4, xref stream for PDF 1.5+) and keeps the version header intact. File size changes by less than 1 percent — just a few bytes per rotated page for the updated /Rotate entries — because no content streams were touched.

Honest limitations

· Rotation is page-level metadata — if your PDF contains rasterized images that are themselves rotated incorrectly within the page (wrong-way text inside a bitmap), rotating the page flips the whole page including other correct content.
· Form field tab order follows page order, not visual orientation — rotating a page with AcroForm fields will not reorder tab stops to follow the new reading direction.
· Digital signatures invalidate because the byte-range hash covers the rotated /Rotate entries; rotating a signed document breaks its signature validity indicator.

Pro tips

Rotate before running OCR for best results

Tesseract and most cloud OCR engines try to auto-detect orientation, but auto-detection fails on pages with large graphics, low text density, or mixed-orientation content. If you know a page is 90 degrees off, rotate it first via this tool before sending to OCR, and the OCR accuracy jumps because the engine no longer has to guess. This is especially helpful for handwritten form fields where Tesseract's orientation heuristic struggles. Rotation via /Rotate metadata is interpreted correctly by modern OCR engines, which read the effective orientation before character recognition.

Use 180-degree rotation to rescue upside-down scans

When a scanner feeds a stack upside-down (common with stapled documents fed by the unstapled edge), every page comes out 180 degrees rotated. Rather than re-scanning, select all pages and apply 180-degree rotation — pdf-lib handles it in under a second regardless of page count. The output is indistinguishable from a correctly-fed scan. If the scanner's auto-rotate feature partially corrected some pages (making them random mixes of 0 and 180), you will need to audit and rotate per-page which is still faster than re-feeding 60 pages through the ADF.

Verify by opening in Preview or Adobe Reader, not your browser

Browser PDF viewers (Chrome's PDFium, Firefox's pdf.js) sometimes cache rotation state aggressively and show pre-rotation orientation until you hard-refresh. After rotating, open the output in Adobe Reader, macOS Preview, or SumatraPDF to confirm the /Rotate values are correctly applied before sending the file onward. This five-second verification step has saved more than one embarrassing re-send to a client who opened the file in a different viewer that respected the metadata correctly.

Frequently asked questions

Does rotating reduce file quality the way re-saving a JPEG does?

No. Rotation in PDF is a pure metadata operation — we change one number on each page dictionary (the /Rotate value from 0 to 90, for example) and save. No images are re-encoded, no fonts are re-embedded, no content streams are touched. The output is bit-identical to the input except for a few dozen bytes of rotation metadata. This is fundamentally different from rotating a JPEG, where decoding and re-encoding always introduces generation loss. You can rotate a PDF ten times without quality degradation, whereas ten JPEG rotations visibly blur the image.

Why does my page display rotated in Chrome but normal in Adobe Reader?

Different PDF viewers handle the /Rotate attribute slightly differently when it combines with content-stream-level rotation matrices. Chrome's PDFium respects /Rotate strictly; older versions of Adobe Reader compose /Rotate with matrix rotations inside the content stream, which can cancel out. The fix is to rotate specifically to match the intended display: if Chrome shows sideways but you want portrait, rotate 90 more degrees in this tool. If inconsistencies persist across viewers, the source PDF likely has a content-stream rotation matrix that conflicts with /Rotate — in that case the underlying page needs re-authoring, not just metadata rotation.

Can I rotate only specific pages instead of the whole document?

Yes. The tool accepts per-page rotation rules via a range selector: 'pages 1, 5, 12-15 rotate 90 clockwise; page 20 rotate 180' and so on. Internally we iterate the page indices and apply the requested delta to each one's /Rotate attribute independently. Pages not mentioned in any rule keep their current rotation unchanged. This is how you fix three misfed scanner pages in a 40-page document without touching the 37 pages that are already correct — a common scenario and exactly what metadata-only rotation is best at.

Will rotating affect the OCR text layer if my PDF has one?

The text layer persists because we do not touch content streams. However, the /Rotate attribute also affects how selection tools render the text layer — rotating 90 degrees means text that was horizontal is now logically vertical, and selecting with a mouse drag works in the rotated direction. The underlying glyph positions in the content stream do not change, so copy-paste produces the same text as before rotation. OCR engines that re-process a rotated document read the effective orientation and apply the text layer with the rotation honored, so subsequent text extraction works correctly.

Is my file uploaded anywhere during rotation?

No. Rotation happens in your browser tab with pdf-lib.js parsing the PDF into JavaScript heap memory, updating the /Rotate attribute on each target page, and saving back to a Blob that downloads locally. No fetch, no websocket, no service worker. You can verify by opening the browser's devtools Network tab while rotating — zero outbound requests will contain file contents. This matters when the PDF holds medical records subject to HIPAA, client financials, or legal documents under attorney-client privilege that cannot be shared with SaaS tools without creating a compliance problem.

Rotation is often a quick fix between other steps. After merging mixed-orientation scans with pdf-merger, rotate-pdf normalizes the combined output before it gets compressed by pdf-compressor for email delivery. When scanned images come in via images-to-pdf, rotation fixes any that were captured in landscape when they should be portrait. For pages that were correctly oriented in the original but need a different orientation for a specific audience (print shop vs screen reader), rotate is the lightweight non-destructive answer. If a digital signature needs to remain valid, rotate before signing, not after — rotating a signed document invalidates the signature.