GUIDE · HOW-TO

How to Redact a PDF (and Check That It Worked)

Published July 22, 2026 · ~7 min read

You have a PDF that has to go out, and a few things in it that can't. A client's Social Security number, a home address, a witness name, a bank account. So you open the file, draw a black rectangle over each one, save, and send.

That is how most redaction failures start. Learning how to redact a PDF is really two lessons: which tools remove the text, and which ones only cover it up. They look identical on screen. They are not the same file.

The one rule: covering text is not removing text

A PDF stores two different things — the instructions for what to draw on the page, and the text itself. A black rectangle is a drawing instruction. Adding one doesn't touch the characters underneath. Any PDF reader can still select them, copy them, and search them.

The U.S. Court of Federal Claims puts it about as plainly as a court can. In its PDF file redaction best practices, it warns that the highlighter function "creates a black box over the data, but it merely hides the data," and that "Anyone can copy the black box, paste it into a word-processing document, and the information under the black box appears."

That warning exists because this failure keeps happening — to law firms, agencies, and courts with real secrecy obligations and careful people. The carelessness is never about secrecy. It's about one specific thing: the difference between drawing and deleting.

How to redact a PDF in the tools you already have

Apple Preview (macOS)

Preview is where most Mac users start, and it has both a right way and a wrong way sitting in the same toolbar.

The right way is the Redact tool — Tools > Redact, or the Redact button in the Markup toolbar. Apple's own Preview guide describes it as: "Select text to permanently remove it from view. You can change the redaction as you edit, but once you close the document, the redaction becomes permanent" (Apple Support). Macworld notes the tool arrived in Preview with macOS Big Sur in 2020 — so if you're on something older, Preview has no true redaction at all.

The wrong way is anything else that produces a black shape: the filled rectangle in Shapes, the highlighter set to a dark color, a black pen scribble. Those are annotations. They sit on top of live text.

Two caveats. Redaction in Preview can't be undone, so work on a duplicate — though the original stays recoverable on your own Mac via File > Revert To or Time Machine. And the Redact tool handles page content, not document metadata. Check that separately (see the checklist below).

Adobe Acrobat Pro

Acrobat has real redaction, and one very common way to get it wrong: it's a two-step process, and step one does nothing.

You mark content for redaction, then you apply it. Adobe's own tutorial spells out the order — "To redact information, first mark the items for removal and then apply the redactions" — followed by a separate prompt to "sanitize and remove hidden information," after which "redactions and hidden information are permanently removed and saved to a new file" (Adobe).

Mark and send without applying, and you've shipped a file with red boxes and intact text. One more thing to know before you plan around Acrobat: redaction is an Acrobat Pro feature. The free Reader annotates; it doesn't redact.

Microsoft Word

Word has no built-in redaction tool, and the workarounds people invent are the dangerous ones — black highlighting, white text, a black shape dropped on top. All cosmetic. Export to PDF and the characters come along for the ride.

Document Inspector doesn't rescue you here. It strips metadata, comments and tracked changes, but Microsoft states plainly that it "does not detect objects that are covered by other objects," and "cannot detect text that was hidden by other methods (for example, white text on a white background)" (Microsoft).

The only honest Word workflow: delete the sensitive text in the source document, type [REDACTED] in its place, run Document Inspector for metadata, then export to PDF.

Print, black marker, scan

Old-fashioned and, for scanned or photocopied material, endorsed by the court guidance above: "print the document, black out the text to be redacted with a black marker and scan the paper back into PDF format."

It works because the scan is an image — there's no text layer left to recover. The costs are real: you lose searchable text and accessibility, the file gets bigger, and thin or streaky marker coverage can still be legible in a bright scan or pulled out by OCR. Go over it twice, then look hard at the scan before you send it.

One thing this is not: "Print to PDF." Saving to PDF from a print dialog flattens annotations into the page, but it commonly keeps the text as text. Don't assume it removed anything — test it.

The 60-second check on your own file

Do this on the file you are about to send, not the one you edited.

1. Copy-paste test. Drag-select across the blacked-out area, copy, paste into a plain text editor. If anything appears, the redaction is fake.
2. Search test. Cmd/Ctrl+F for a name, a street, the last four digits of an account. Then select all, copy the whole page, and read what you get.
3. Metadata test. Open document properties. Title, author, subject, and keywords carry names surprisingly often — as does the filename itself.
4. Hidden-content test. Look for attachments, comments, bookmarks, and layers. Each is a place text can survive out of view.
5. Pixel test. Zoom to 400% along the edges of each box. A slightly transparent or slightly misplaced box leaves a readable sliver.

These five catch the common failures. Be clear about what they don't catch: they test whether your removals worked. They can't tell you about the date of birth on page 84 that you never noticed.

The failure that checking can't fix

Fake redaction is a tooling problem, and now you can spot it. Missing something is an attention problem — and it gets worse with page count, deadline pressure, and the fourth document of the afternoon.

That's the gap RedactWorks is built for. It scans the document and proposes the PII to redact from your PDF — names, SSNs, financial and medical identifiers, plus rules you define. You review and approve every finding; nothing is removed without your sign-off. Approved items have their text layer removed and metadata stripped, so nothing is recoverable from the output. Then it re-scans the finished file and, if sensitive data still remains, blocks delivery and quarantines it. High-risk categories are locked on by default, and an override takes an explicit, logged acknowledgment.

No software can promise a perfect outcome, and the judgment calls stay yours. What a verify pass gives you is a second, independent check that refuses to hand over a file that fails it. See how the workflow runs, or read what we do and don't do with your files on the trust page.

Frequently asked questions