Remove Watermark from PDF

Q: Which mode should I use?

Try matching the watermark text first if you know it exactly. If that finds nothing, use the brightness-lightening mode for a scanned page, or the region-cover mode for a logo confined to one spot.

100% processed in your browser. A "watermark" means something different depending on how the PDF was made, so this tool offers three approaches -- pick whichever matches your file.

Drop a PDF here
or click to browse

Best when the watermark is real, selectable text (added by a tool, not baked into a scan) -- the exact watermark text is found and deleted from each page, leaving everything else, including your real text, completely untouched and still selectable.

Watermark text to remove

Matching isn't case-sensitive, and only needs to match part of the text drawn at that spot.

Works on any PDF, text-based or scanned -- draw a box over the watermark below, and that exact region gets covered in white on every page, at the same position. This doesn't know what's underneath, so avoid boxing over real content you want to keep.

No region selected yet.

Best for a faint, page-wide, or diagonal watermark baked into a scanned page -- any pixel lighter than the threshold below is pushed to pure white, erasing a light watermark while leaving darker real content alone.

Brightness threshold (200 / 255)

⚠️ Limit: File must be under 25MB.

Frequently Asked Questions

Is my PDF uploaded to your server?

No. Everything happens inside your browser -- your file is never uploaded anywhere.

Which mode should I use?

Try "Match the watermark text" first if you know the exact watermark text (like "DRAFT" or "CONFIDENTIAL") -- it's the cleanest option, since it deletes only that specific text and leaves everything else, including your real content, fully intact and still selectable. If that finds nothing (common for a scanned document, where there's no real text to search), use "Lighten a scanned page" for a faint, page-wide watermark, or "Cover a region" for a logo or stamp confined to one spot -- either works regardless of whether the PDF has real text or not.

Why didn't "Match the watermark text" find anything?

Either the watermark isn't real text (it's baked into a scanned image -- try one of the other two modes), or it's real text drawn with an unusual embedded font that doesn't decode the way this tool expects. This tool never guesses -- it tells you plainly when nothing was found rather than silently producing an unchanged file.

Does this use AI to detect or remove watermarks?

No. Each mode uses a simple, predictable method -- exact text matching, a boxed region, or a brightness cutoff -- not AI-based image understanding. That also means none of them can tell "watermark" apart from "real content" on your behalf; that judgment is yours.

Can I remove more than one watermark, or a watermark plus other edits?

Run the tool again on the result for a second watermark, or a different mode. Each pass produces a new PDF you can feed straight back in.