PDF Page Rotator

PDF page rotator: fix sideways and upside-down pages, one page or all, entirely in your browser

Scanners produce crooked orientation constantly. A document goes through the feeder the wrong way up, a landscape spreadsheet lands in a portrait batch, someone rotates a page in a viewer and the change turns out to be a display setting that nobody else sees. This tool fixes the orientation permanently in the file: pick the pages, turn them 90° clockwise, 90° anticlockwise or 180°, and download a corrected PDF.

Your file is not uploaded. The PDF is read into memory with the browser's own file API and rewritten by a JavaScript PDF library running on this page. Nothing leaves your machine, there is no server processing step, and no copy exists anywhere for anyone to delete later. That matters for the documents people actually need to rotate — scanned contracts, medical records, tax paperwork, HR files.

How to use it

• Drag a PDF onto the drop zone, or use Choose File. The filename, size and page count appear.
• Click the numbered page buttons to select the pages you want to turn. Selected pages highlight. Select All and Deselect All handle the whole-document case.
• Press 90° CW, 90° CCW or 180°. A badge appears on each affected page showing the total rotation queued for it.
• Change the selection and rotate again if different pages need different amounts. Rotations accumulate per page, so pressing 90° CW twice on a page shows 180°.
• Apply Rotation, then download the result. The output is named after your original with _rotated appended.

Reset Rotations clears all queued turns without discarding the loaded file; Reset clears the file entirely. Nothing is written until you press Apply, so you can queue, clear and re-queue freely.

One honest limitation: the page picker shows numbered buttons, not thumbnails. There is no preview of what each page looks like, so you need to know which page numbers are wrong — open the PDF in a viewer alongside this tool and note them first. For a document where every page is wrong the same way, which is the usual scanner case, Select All plus one button press is the whole job.

Which way is 90° clockwise?

People get this wrong constantly, so here is the reference by symptom rather than by geometry:

If you queue the wrong direction, press the opposite one to cancel it out, or hit Reset Rotations and start again. Because the tool tracks a cumulative angle per page rather than a history, 90° CW followed by 90° CCW returns cleanly to zero.

The thing that confuses everyone: rotation is a flag, not rotated content

This is worth understanding because it explains almost every strange result people get with PDF rotation.

A PDF page has a /Rotate entry in its dictionary, holding 0, 90, 180 or 270. It is an instruction to the viewer: display this page turned by this much. The page's actual content — the text positions, the scanned image, the vector drawing — is untouched by it. A landscape scan can therefore exist in two completely different forms that look identical on screen:

• A genuinely landscape page: the media box is wider than it is tall, and the content is drawn that way.
• A portrait page whose content is drawn sideways, with /Rotate 90 telling the viewer to turn it.

This tool changes the flag. It reads each selected page's current /Rotate value and writes back that value plus your rotation, so the operation is additive to whatever the file already had — a page already at 90 that you turn another 90 ends up at 180, which is what you would expect and is not always what other tools do.

The consequences of working at the flag level are all good ones:

• It is completely lossless. No image is re-encoded, no text is re-rendered, no quality is lost, and the file size barely moves. Rotating a 40 MB scan does not produce a 55 MB scan.
• Text stays selectable and searchable. The text layer is not disturbed at all, so search, copy-paste and any OCR layer already present survive intact.
• It is instant, even on large files. The work is proportional to document structure, not to page content.
• Everything else in the document is preserved — bookmarks, form fields, annotations, links, metadata. The original document is loaded, the rotation entries are edited, and it is saved back.

And one caveat, which is where the confusion comes from: a small number of badly-written viewers, older printers and some naive text-extraction scripts ignore /Rotate and render or read the raw content. On those, a page you rotated here will still appear sideways. This is not a defect in the file — it is the correct, standard way to express page rotation, and every mainstream viewer honours it. If you hit such a tool, the fix is to rasterise the page and rebuild it, which is a fundamentally lossy operation and a different job.

Rotation versus what your viewer's rotate button does

This is the most common reason people go looking for a rotation tool at all. In most PDF readers, the rotate control on the toolbar changes your view and is not saved to the file. You rotate the page, it looks right, you email it, and the recipient sees it sideways again. Some readers offer a separate "rotate pages" command that does write to the file, usually buried in a page-organisation panel and often behind a paid tier.

Common jobs

The duplex case deserves a note because it is so common and so confusing. A duplex scanner that flips pages on the short edge when the document was printed for long-edge binding produces a file where every back side is upside down. The symptom is unmistakable: odd pages correct, even pages 180° out. Select the even ones, press 180°, done — and if it happens repeatedly, change the flip-edge setting on the scanner rather than fixing every batch here.

What page rotation cannot fix

Rotation is defined in the PDF format only in 90° increments, so a whole family of scanning problems look like rotation problems and are not:

The last one is worth watching for. Rotating a page swaps its effective width and height for display purposes, so a page whose content was already positioned oddly relative to its media box can end up looking shifted. The content has not moved — the frame around it has turned. If a rotated page prints with a strange margin, the media box, not the rotation, is what is wrong.

Doing the same thing on the command line

If you are rotating hundreds of files rather than one, this is a scripting job. Two tools do exactly what this page does, at the same flag level and equally losslessly:

• qpdf in.pdf out.pdf --rotate=+90:1-z rotates every page 90° clockwise. The + means relative, matching this tool's additive behaviour; drop it and the angle is set absolutely. Replace 1-z with a page range such as 2,4,6 or 3-8.
• pdftk in.pdf cat 1-endeast output out.pdf does the same in pdftk's vocabulary, where east is 90° clockwise, west anticlockwise and south 180°.

For a handful of files, the browser is faster than installing either. For a recurring batch out of a document scanner, the command line is the right answer — and the better answer still is to fix the scanner's feed orientation setting so the batch arrives correct.

Rotation and OCR

If you are about to run OCR over a scan, rotate first. OCR engines read horizontal text; a sideways page produces either nothing or convincing gibberish. Most OCR tooling honours the page rotation flag when it rasterises, so correcting orientation here before the OCR step is usually enough. If your OCR tool produces garbage anyway, that is a strong sign it belongs to the minority that ignores /Rotate, and you would need it rasterised.

Questions

• Is my file uploaded? No. It is read and rewritten in the browser tab. You can verify by disconnecting from the network after the page loads — the tool keeps working.
• Does rotating reduce quality? No. The page content is not touched; only the display-rotation value changes.
• Will the file get bigger? Essentially no. The rewritten file is the same document with a few dictionary entries changed.
• Can I rotate by an arbitrary angle to fix a crooked scan? No. PDF page rotation is defined only in 90° steps, and this tool offers exactly those. Straightening a skewed scan means rotating the image content itself — a rasterising, lossy operation that no page-rotation tool performs.
• It says the PDF may be corrupted or encrypted. Password-protected files cannot be parsed without the password. Remove the protection first, then rotate.
• Are my bookmarks and form fields still there? Yes. The original document is edited in place rather than rebuilt, so its structure is preserved.
• Nothing happens when I press Apply. At least one page needs a non-zero queued rotation. Selecting pages is not enough — you also have to press one of the rotation buttons.
• Can I change the page order at the same time? Not here; this tool only changes orientation. Sequence is the PDF Page Reorderer's job.