Turn individual pages or the whole document and download a copy that stays that way. The rotation is stored in the file rather than applied by your viewer, so it travels with the document.
Runs in your browser. Nothing is uploaded.
Every page in a PDF carries an optional entry recording how it should be displayed, in ninety degree steps. Rotating here changes that number and saves the document. No content is moved, no image is decoded and nothing is re-rendered: the page description is identical afterwards and only the instruction about which way up to draw it has changed. Viewers, printers and browsers all honour it, which is why the result looks right everywhere rather than only where you fixed it.
The measurements make the point better than the explanation does. Turning all twelve pages of a 50,563 byte text document produced a file 15 bytes larger, in 17 milliseconds. Doing the same to a 10,324,119 byte scan added 14 bytes and took 33 milliseconds. A ten megabyte document and a fifty kilobyte one cost almost exactly the same, because the work has nothing to do with what is on the pages.
Choose a PDF
The same questions a security review would ask, answered plainly.
Rotate the whole document in one press, or turn only the three sheets that went through the scanner sideways. Each page keeps its own setting.
Text stays selectable and images keep their original resolution, because the page content is untouched and only its display angle changes.
The document is modified in place rather than rebuilt, so a fillable form is still fillable afterwards and the outline tree is still there. Not every PDF operation can say that.
A 10 MB scan took 33 milliseconds, barely more than a 50 KB document. Cost here does not follow file size, because no pixels are involved.
There is no upload step and no server to receive one. The page loads once, then all work on your PDFs happens in local JavaScript.
No daily limits, no per-file ceilings imposed by a server tier, no watermarks, and no account. The practical limit is your own device's memory.
A small file and a large one, costing almost exactly the same.
Twelve pages, all rotated ninety degrees.
Input
12-page text PDF, 50,563 bytes
Output
50,578 bytes, 17 ms
Settings: All Right, applied to every page
Fifteen bytes larger across twelve pages, which is the rotation entry written onto each one. Nothing else in the file changed.
The case that shows rotation does not care what is on the page.
Input
12-page scan, 10,324,119 bytes
Output
10,324,133 bytes, 33 ms
Settings: All Right, applied to every page
Fourteen bytes added to a ten megabyte document, in about twice the time the small file took. The scanned images were never decoded, so their size is irrelevant to the work.
Real files, real numbers, and the machine they were run on.
| Scenario | Input | Output | Time | Notes |
|---|---|---|---|---|
| all 12 pages of a text PDF turned 90 degrees | 50,563 bytes | 50,578 bytes | 17 ms | 15 bytes added |
| all 12 pages of a 10 MB scan turned 90 degrees | 10,324,119 bytes | 10,324,133 bytes | 33 ms | 14 bytes added |
How this was measured: Measured by driving this page in headless Chrome 152 on an Apple M4 Pro running macOS 15.5, on 2026-09-17. The fixture is handed to the page's own file input, the page's own button is pressed, and the figure recorded is the size of the file the page produces. Timings come from a warm page on a fast machine with no network involved, so treat them as a floor rather than a typical result.
A sheet fed in the wrong way round produces a sideways page in an otherwise correct document. Turning that one page is quicker than rescanning the stack.
A printer will honour the stored rotation, so correcting it in the file avoids the guesswork of landscape settings and manual feed orientation.
A contract where page four arrives upside down looks careless. Fixing it locally takes seconds and the document never goes near a third party.
A sideways page is genuinely hard to read on a small screen, since the reader scales it to fit the wrong dimension. Correcting the file fixes it on every device at once.
Archives and submission portals often reject or flag incorrectly oriented pages. Correcting them before submission avoids a rejection days later.
Rotation runs entirely in the browser once the page has loaded, so a document can be tidied up with no connection at all.
PDF records orientation as 0, 90, 180 or 270 degrees, so a page scanned at a slight angle cannot be straightened here. Correcting a few degrees of skew needs software that re-renders the page.
Rotation changes how a page is displayed, not where the ink sits on it. A page whose text was drawn sideways into an upright page will still read sideways after rotating, because the page itself was never the problem.
Turning a page swaps how tall and wide it appears without changing its dimensions. A landscape page rotated into portrait is still a landscape page being displayed upright.
No. The file is opened by code running in this tab, one number per page is changed, and a new copy is written to your downloads. Nothing is transmitted, and the operation works with the machine disconnected from the network entirely.
Yes. The angle is written into the document rather than applied by a viewer, so anyone who opens it sees the corrected orientation. This is the difference between rotating here and rotating in a reader, where the change is often just a display preference that is not saved.
Yes, each page has its own controls and its own stored angle. Mixed orientations in one document are fine, which is the usual situation when a couple of sheets went through a scanner the wrong way.
Neither in any meaningful way. Nothing is re-rendered, so text stays selectable and images keep their resolution. Twelve pages of a 10 MB scan grew by fourteen bytes, which is the rotation entries themselves.
No. The format only allows quarter turns, and this writes one of those. Deskewing a page that is a few degrees off needs a tool that redraws the content, which necessarily rasterises it.
Yes, and that is not true of every PDF operation. Rotation modifies the document in place rather than copying its pages into a new one, so form fields, the bookmark tree and anything else held at document level all survive. Merging or reordering the same file would drop them.
The rotations add up, so two turns to the right give you 180 degrees and four bring you back to where you started. The value is normalised, so you cannot end up with an invalid angle by pressing a button too many times.
None imposed here. Rotation is unusually light work, so the practical ceiling is simply holding the document in memory while it is rewritten. Very large scans are fine because their images are never decoded.
No. A corrected copy is written to your downloads and the file on your disk is left exactly as it was, so getting the direction wrong costs you one download rather than your source document.
Yes, after the first visit. Reading a page tree and writing one number onto each sheet needs no outside help, so the network is irrelevant from the moment the tab finishes loading.
There is no limit imposed by us, because there is no server tier to enforce one. The real constraint is your device's available memory, since the whole file is held in RAM while it is processed. Very large files are slower on phones than on a laptop.
No. There is no sign-up, no email, and no usage tracking tied to an identity. The tool works the first time you open it.