Redact PDF
Censor sensitive text in your PDF
Click or drop a PDF file hereChoose a PDF file
About Redact PDF
Redact PDF finds the words or phrases you list, one per line, and, with the matching options checked, every email address, phone number, and card number in the text of your PDF and in its form fields and comments. The search ignores case, spaces, line breaks, accents, and hyphens. Each page with a match is rebuilt as a picture with the matches blacked out in the pixels, plus an invisible text layer for the other words, so the hidden text is gone from the file. Each black box is fitted to the glyphs of the matched letters, so the words beside them stay readable. The finished PDF is searched again and withheld if a term can still be found. Scanned pages have no text to search: use OCR PDF first.
How to redact text in a PDF
- 1Choose the PDF; a file that needs a password to open is refused at once.
- 2Type each word or phrase on its own line in "Text to redact (one word or phrase per line)", and check "Email addresses", "Phone numbers", or "Card numbers" if needed.
- 3Click "Redact PDF" and follow the page-by-page progress.
- 4Read the summary of occurrences per page, then click "Download" to save redacted.pdf.
Formats and limits
- Input
- PDF with a text layer
- Automatic patterns
- Email addresses; phone numbers of 9 to 15 digits; card numbers of 13 to 19 digits that pass the Luhn check
- Pages with a match
- Replaced by a picture with black boxes and invisible text for the remaining words, Arabic included
- Black boxes
- The matched letters' glyphs plus 2% of the font size or one pixel, whichever is larger, and half the outline width for outlined text; 15% of the font size on each side for invisible text over a scan (an OCR layer) or where a line's glyphs cannot be tied to its text, or the whole line
- Pages without a match
- Copied, minus stamps, attachments, media, and links that jump to a redacted page, run a script, or open another file; the redacted file keeps no bookmarks, document properties, or fillable form fields
- On iPhone and iPad
- A matched page that fails or is not drawn within 20 seconds is drawn by our PDF service, for PDFs up to 44 MB, up to 300 pages per hour and 1,000 pages per day per network
- Result
- redacted.pdf
Where your file is processed
On a computer or an Android device, searching, drawing, and blacking out all happen in your browser, and the PDF is not uploaded. On an iPhone or iPad, a page with a match that the device fails to draw, or does not draw within 20 seconds, is drawn by our own PDF service (pdf-tools, not a third party): the PDF is sent there and then deleted, and the black boxes are still applied in your browser. A page whose text cannot be trusted (some PDFs, often Arabic ones, give characters the wrong letters) is also read by OCR: in your browser with Tesseract.js, whose engine and language data are downloaded from cdn.jsdelivr.net the first time, or, on an iPhone or iPad, by our own OCR service, which receives the PDF and then deletes it. The page tells you when this happens.
Frequently asked questions
Is the text really removed, not just covered?
Yes. A matched page is replaced by a picture of itself with the matches blacked out in the pixels, so no text sits under the boxes; only the other words come back as invisible text. The saved file is searched again, and if a term can still be found, no file is given.
Will the rest of a redacted page stay searchable?
Yes, mostly. Words outside the black boxes are written back as invisible, selectable text in reading order, Arabic words included, as the PDF's own text gives them. Not kept: words that touch a box, vertical text, and words with letters the text layer cannot write, such as Greek, Cyrillic, Hebrew, Asian scripts, and some accented letters (ő, ł, ș…); a page whose kept words could spell a term together keeps none. Links and form fields on that page are gone.
Does it find a name written with or without accents?
Yes. Accents, case, spaces, line breaks, and hyphens are ignored, so Muller also finds Müller, and a word split across two lines is found too. This can black out slightly more than you typed, never less, so check the pages listed in the summary.
Do pages without a match stay as they were?
Yes, mostly: their text, images, and ordinary links stay. Stamps, file attachments, and media are removed because their content cannot be checked, as are links that jump to a redacted page, run a script, or open another file. The whole redacted file also loses its bookmarks, document properties, and fillable form fields.
Can it redact a scanned PDF?
No. A scan has no text layer, so nothing can be found and the tool reports that there was no match. Use OCR PDF to get a searchable PDF, then redact that file.
Can it miss a word that is in the PDF?
Rarely. Some PDFs, often Arabic ones, give some of their characters the wrong letters, so their text cannot be searched. Redact PDF spots such pages and also reads them by OCR, up to 80 such pages per redaction (a PDF with more is refused, with the way to split it), then blacks out what OCR finds there; on a computer, the blacked-out page is read again by OCR before the file is given. OCR can misread a letter too, so the summary names the pages read by OCR and any term found nowhere. If a term could still be read in the finished file, no file is given.
Tips and tricks
- ✓Prefer full phrases to short fragments: a short term can match inside other words and turn more pages into pictures than needed.