How to Sign a PDF Without Uploading It (Free, No Sign-In)

A lease that has to go back to the landlord. An offer letter before the reply deadline. A tax form, a consent, a contract where the words still have to be selectable afterward. In all of them the signature is the part that decides whether the document is done, and in most of them the document is the one you would least want a stranger to hold.

The usual route is to hand the PDF to an online signer. That is a reasonable thing to do and an unreasonable thing to do at the same time: reasonable, because signing should be quick; unreasonable, because the signature is the most personal mark you own and it is going onto the most sensitive file you have, and the large signing services take that file onto their servers. Adobe states under its own drop zone that the file is handled by Adobe servers; Smallpdf keeps it for an hour; a subscription service will ask you to sign in before it will let you download what you just produced. The signature and the document both leave your machine.

This guide is about the other way: signing a PDF in the browser so that neither the document nor the signature is ever uploaded. It also settles the question that most signers quietly skip — what the mark you just drew actually is, and when that is enough.

What an electronic signature is, and what it is not

Most of the confusion around "signing a PDF" comes from two things that are not the same thing being called by one name.

An electronic signature is a mark that shows intent to sign. Your name drawn on a canvas, typed in a handwriting font, or a scanned signature pasted onto a page — all of these are electronic signatures in the ordinary sense. The United States ESIGN Act (2000) and the EU eIDAS regulation both hold that a contract or a signature may not be refused legal effect simply because it is electronic. For contracts, leases, offer letters, NDAs, invoices and most HR and commercial documents, this is what the law actually requires, and it is what the tool on this page produces.

A digital signature is something else: a cryptographic object (PDF PAdES, in the PDF world) that carries a certificate, proves who signed, and shows whether the page changed afterward. That is what a regulated filing, a notary, or a document that must survive a tamper challenge needs. The drawn mark on this page does not do that — there is no certificate, no identity proof, and a reader cannot tell whether the page was altered after signing. Where the signature has to be verifiable, a drawing tool is not the tool; you need a signing service with an identity check, and this one is honest that it is not that.

Most people signing a lease or an offer letter are doing the first kind, not the second. The point of being clear about it is to stop over-promising: the download you get is a signed document in the legal sense that matters for everyday papers, and it is not a tamper-evident seal.

Signing without an upload

The whole point is that the file never leaves the machine. The PDF is read by the browser's own file API, the signature is drawn on a canvas in the page, and the signed PDF is assembled in memory before it is handed to the downloads folder. There is no upload request, no processing queue and no server that could keep a copy — which is exactly the inverse of what the cloud signers do, and it is why an NDA, which exists to keep something confidential, is one file you especially do not want a third party holding.

Two things follow that the ranking pages do not mention. First, nothing is retained: close the tab and the signature is gone. It is not written to local storage, a cookie or a cache, deliberately, because a signature saved in browser storage is a signature anyone with that machine can reuse on a document you never saw. Second, the signature is trimmed to the ink before it is embedded, so it sits exactly where you put it instead of inside a transparent box, and it is a few kilobytes regardless of how large you draw it.

Where the signature actually lands

The mark is easy; the placement is where the tools go wrong, and the failure is silent — the signature simply arrives on its side or half clipped, with no error anywhere.

A scanned page is usually stored as landscape pixels plus a /Rotate instruction that tells the reader to turn it. A page prepared for print carries a CropBox smaller than the physical sheet. Most tools place the signature in the page's own coordinate system, so on a page the reader displays sideways the signature is sideways, and on a print-ready page it lands in the strip the CropBox clips away and is invisible. The tool on this page plans the placement against the page as you see it and maps it back through both the rotation and the visible area, so what you drag on screen is where it lands.

Measured, not assumed: on a three-page document whose last page is stored with a 90-degree rotation, the signature is drawn at page-space angles 0, 0 and 90 — upright on screen on every one of the three pages. The three signatures share a single embedded image asset, so stamping a mark on every page does not re-embed the image once per page — the same object is referenced. Because the mark is a transparent PNG, that asset is one image object plus its alpha soft-mask (two objects total), and that is the expected shape, not a leak.

Does signing flatten the document?

Neither. The signature is embedded as an image object in a new content stream on top of the existing page. The original text, fonts, images and vector artwork are carried through untouched, so the words stay selectable and searchable and the file stays a normal PDF any reader opens. In the measurement the original sentence "This lease is executed in two copies" is still in the content stream after the signature is placed — the document was not rasterised into an image, which is the other thing a flattened page quietly does.

The file does grow, because a signature is real bytes: the three-page lease went from 1,587 bytes to 3,188 bytes with the mark on every page, and 2,979 bytes with it on the last page plus the date line. That is the cost of a real embedded mark, and it is the number to check when the output is the size you expected.

Signing every page, and the date line

For a long agreement, the initials case is the one that used to mean twenty-four clicks. The scope control offers the last page, the first page, the page you are looking at, or every page; one placement applied to all pages signs a 24-page agreement in a single action. An optional date line prints today's date under the signature at a size that stays legible even when the mark is small — the measurement shows it landing below the signature on a flat page and above it when there is no room, decided in the displayed space so it reads correctly on a rotated page too.

When not to use it

Password-protected PDFs are refused with a sentence that says so, because a protected file's page content is encrypted and a signature written into it could not be encrypted to match — the result would be a file no reader can open. Remove the password in a PDF reader first, then sign. And where the signature must be verifiable or tamper-evident, this is the wrong tool by design; say so in the document you send rather than let a reader assume a drawn mark is a seal.

The short version

Sign a lease, an offer letter, an invoice or a consent in the browser: draw the signature, place it, and download. The document never leaves your device, the signature is never stored, and the text stays selectable afterward. Know what you are producing — an electronic signature, enough for everyday papers, not a certificate — and check the two things that decide whether it is visible: the rotation and the crop box.