Sign PDF files

What this actually is, before you use it

This page puts a signature onto a PDF page as a mark: ink you drew, a name set in type, or an image you already had. It is not a cryptographic signature. There is no certificate, no signing key, no tamper detection, and nothing here can verify a signature somebody else applied. Saying that first is deliberate, because the industry mostly does not. One popular competitor files this exact operation under the heading digital signature of documents, which is the name of a completely different technology, and people reasonably conclude they have got something they have not.

The distinction matters in exactly one situation and is irrelevant in every other. For the ordinary business of agreeing to things — a lease, a consent form, a school permission slip, a contractor's quote, a power of attorney, an offer letter — what the law generally asks for is that a person intended to sign and that the mark is attached to the record. That is the shape of the US ESIGN Act and UETA, and of the simple electronic signature tier in the EU's eIDAS. A drawn or typed mark on the page satisfies it, and it is precisely what happens every time somebody prints a document, signs it in biro and scans it back — only without the printer, the scanner and the loss of quality. Where it is not enough is when a specific counterparty has told you it is not: a court filing system demanding a qualified signature, a bank requiring a certificate, a procurement portal with its own validator. If somebody has said that to you, close this tab and use their tool. You should learn that before you send the file, not after it bounces.

PDF, in brief

PDF

Full name	Portable Document Format
Extension	.pdf
Format type	Fixed-layout document
MIME type	application/pdf

Sign PDF files without handing the document to anybody

All of it happens here. When you first hand something in, two programs come down from this domain — a page renderer and a PDF writer — and they are the same two files for every visitor regardless of what anyone is signing. Once they have arrived the connection has nothing further to say. The document is opened off your own disk, edited in memory and written back out to it; your signature makes the same short journey and no other.

That is worth more here than on almost any other tool on this site, and the reason is not abstract. Documents that need signing are leases, settlements, offers, medical consents and powers of attorney. They name people, they carry addresses and amounts, and the other party to the agreement never consented to a third company keeping a copy. Nor did the specimen of your handwriting you just drew, which is a durable piece of identity you cannot rotate like a password. Assertions of this kind are worth nothing until somebody tests one, and testing this one takes a minute: open the network panel, sign something, and read what is listed. Programs arrive. Nothing departs.

Three ways to make the signature, and why the typed one is not cursive

Draw it with a finger or a mouse. This is the one that is genuinely your hand, and on a phone or a tablet it is close to signing on paper. Whatever you draw is trimmed to the edges of the ink afterwards, so the width control scales the stroke and not the empty space around it — the common annoyance where a short signature drawn in the corner of the box ends up tiny on the page simply does not arise.

Type your name and it is set in Times Italic. Not in a handwriting face, and the reason is measured rather than argued. Rasterizing a name in a browser font means asking the operating system for something like cursive or Segoe Script, and what comes back depends entirely on which fonts that particular machine happens to have. Tested here across two engines on the same Mac: the generic cursive keyword, Segoe Script, Dancing Script and Lucida Handwriting all silently fell through to the plain default face, while Brush Script MT, Snell Roundhand and Bradley Hand each resolved to something different — four different widths for the same name at the same size. The browser reports success in every one of those cases, and the API you would use to check availability answers yes for a font that does not exist. Times Italic is one of the fourteen fonts every PDF reader is required to have, its metrics are identical on every machine, it costs nothing to download, and it comes out of the other end as real selectable text rather than a picture. It looks like italic type because it is italic type. If you want your own hand, draw it — that option is right there.

Hand in an image if you already have a scan or a photo of your signature. A pasted screenshot works the same as a file, because that is how signature images usually travel. One thing to know: a JPEG has no transparency at all, so a photo of ink on white paper stamps an opaque white rectangle over whatever it sits on, usually the printed signature line it was aimed at. There is a switch to knock the light background out, with a preview and with the surviving proportion of the image printed beside it, and it does what it says and nothing cleverer: it measures brightness. Ink on white paper lifts cleanly, ink on gray or cream or lined paper brings the paper with it. That is why it is a switch you throw and look at rather than a treatment applied on your behalf.

Placing it, and how you know it landed

Tap or click the page where the signature should go and it appears centered on that point; drag it afterwards to nudge it. Tapping first rather than dragging first is not a mobile concession, it is the better interaction with a mouse too — a drag around a target is always a few pixels off and a tap never is. Width is a slider with the value written out in both inches and points, every placement is listed so it can be removed individually, and one button puts the same mark on every page for the documents that want a mark on every sheet.

Then the tool checks its own work, which no competitor does. After the file is written it is opened again, each page that received a mark is redrawn, and the quantity of dark pixels inside the exact rectangle you chose is compared against the same rectangle in the original. More ink than before means the mark is genuinely in the finished document. No change means it is not, and you are told so plainly rather than shown a green tick. This check exists because the failure it catches is invisible to every other kind: a signature stamped into the margin produces a file that opens perfectly, has the right page count and contains an image, and passes every structural test anyone would think to run. The finished document is then shown to you in the browser's own viewer, because a download button with no preview forces a blind download.

What happens to the rest of the document

Nothing, if the file is not encrypted. The document is edited in place rather than rebuilt, so its text stays selectable, its fonts stay embedded, and its form fields, bookmarks, links and metadata are all left exactly as they were. Only the pages that receive a mark are touched at all, which is why signing the last page of a four-hundred-page contract costs about the same as signing a four-page one. The one thing deliberately suppressed is our own fingerprint: the PDF library would otherwise overwrite the producer string with its own name and stamp a fresh modification date onto somebody else's agreement, and it does neither here.

An encrypted source takes a different path and gives up more, and you are told which before you run it. If the file needs a password to open there is a field for it; if it carries only an owner password — the sort that restricts printing rather than opening — nothing is asked of you at all. Either way the signed copy comes back without the protection, because writing a modified copy of an encrypted document reliably means rebuilding it, and the encryption dictionary and the form fields cannot come along for the ride. Put the protection back deliberately afterwards with password protect PDF, which also lets you choose the algorithm rather than inheriting whatever the original used.

The same signature on whatever you happen to be holding

Signing is one of the few document jobs where people switch devices mid-task. The lease arrives by email on a phone, gets read properly on a laptop, and has to go back before the end of the day from whichever one is nearest. This page behaves identically on all of them, because it is the same code doing the same arithmetic — a mark placed at the same point on a Windows desktop, an iPad and a ten-year-old Linux box comes out at the same coordinates, the same width and the same darkness. That is worth stating plainly, because it is the one thing a per-device tool cannot promise.

Two things here have no built-in equivalent anywhere. One button puts the same mark on every page, which is what initialling a twenty-page contract actually requires. And after the file is written the tool reopens it and checks that the ink really landed inside the rectangle you chose, which is the failure no other kind of test can see — a signature stamped into the margin produces a file that opens perfectly and passes everything else.

Apple hardware does ship a signing feature and it would be strange not to say so. Markup on an iPhone or iPad, and Preview on a Mac, both keep a reusable signature and neither uploads anything. If the file is already on that device and you are content to keep your signature stored on it, they are a good answer. They are also tied to that one machine: the signature saved into Preview is not the one on your phone, and on a borrowed or work-managed Mac saving it at all may not be what you want. Neither shows you what the finished file carries, and neither of them exists on Android at all — nothing in that operating system signs, and the two preinstalled Google apps that draw on a page both put the marked copy in Drive rather than back where the file was, which is a different bargain than it looks like. Windows has no built-in signing either; the standard advice there is a download and an account.

Jobs that belong on a different page

It does not remove a signature from a PDF, and that direction will not be built. The nearest legitimate version of that request — covering a signature that should not be visible in a copy going elsewhere — is redaction, and redacting a PDF does it properly by rebuilding the page rather than laying a rectangle over it.

It does not fill in form fields, and it does not flatten them. If your document is a real AcroForm, fill it in whatever you normally use, save it, and sign the saved copy here — the fields survive intact through an unencrypted file. It does not assemble documents either: if the thing you need to sign arrived as several files, merging the PDFs first gives you one document with one signature instead of three, and if you only need to return the signed page, splitting the PDF is usually the safer thing to send. Turning a page into a PNG is the right tool when somebody wants a picture of the signed page rather than the document, and stripping the metadata is worth a moment before a signed file leaves the building, because the author field frequently still names whoever drafted it.