How-To 8 min read 2 October 2026

How to Redact a PDF Properly: The Complete Guide

Redaction is not adding a black box — it is permanently removing data from the file. This guide explains why visual-only covering fails, how true PDF redaction works technically, the correct workflow in both online and desktop tools, and how to verify the result is actually safe.

Redaction is not adding a black box. That is the single most important thing to understand, and the single most commonly misunderstood fact about PDF redaction. A black or white rectangle placed on top of sensitive text is a whiteout overlay — it changes the visual appearance of the page without touching the underlying content stream. The text is still there. Copy it, extract it, or delete the rectangle, and it reappears. True redaction permanently removes the underlying data from the PDF's content stream. After correct redaction, the sensitive content does not exist in the file — it cannot be recovered by removing a layer, copying text, or analysing the raw bytes. The redacted area is replaced with a flat black or coloured mark that contains no information. The distinction matters in practice because failures to redact correctly have had serious consequences. The Manafort filing in 2019 exposed the details of a key prosecution allegation within hours of the document's release, because lawyers applied black boxes as visual overlays. The AT&T NSA surveillance filing in 2006 exposed classified evidence for the same reason. These were not edge cases — they were the natural result of a technically incorrect approach.

How true redaction works technically

In the PDF specification, true redaction is a two-stage process: Stage 1: Mark for redaction — A /Redact annotation is placed over the content to be removed. This is a native PDF annotation type introduced in PDF 1.7. The annotation stores the coordinates of the area to be redacted and the visual appearance of the mark that will replace it (typically a solid black rectangle, but the colour and overlay text are configurable). At this stage, the underlying content is still present — the annotation is a pending instruction, not a completed action. Stage 2: Apply redactions — The "apply" step executes the redaction: the tool rewrites the PDF's content stream, removing all text, path, and image operators that fall within the marked rectangles. The /Redact annotations are replaced by their visual appearance marks, which become permanent parts of the page content. The underlying data is gone. The critical failure mode is stopping at Stage 1 — placing visual marks without applying the redactions. This produces a PDF where the marks are present but the data remains. Many cases of failed redaction involve this exact error: someone placed black annotations, confirmed they could not read the text, and distributed the document without realising the data was still there. After applying redactions, the tool must also remove related data from elsewhere in the file: the text layer (for searchable text), embedded metadata that might reference the redacted content, bookmarks pointing to redacted areas, and comments or annotations that contain the sensitive text.

Step-by-step: redacting in PDFTheGuide

PDFTheGuide's redaction tool performs true content removal in-browser — your file never leaves your device. The process: 1. Open the PDF — Drag your document onto the editor or click to browse. Files up to 50 MB are supported. 2. Select the Redact tool — Click the Redact button in the toolbar. The cursor changes to a crosshair. 3. Draw redaction rectangles — Click and drag to draw rectangles over each piece of content to be removed. Draw one rectangle per area. For multi-line text, draw a rectangle that covers all lines. For images, draw a rectangle over the entire image. 4. Download the redacted PDF — Click the Download button. The tool applies the redactions during the PDF generation step using pdf-lib — the content stream is rewritten before the file is produced, so the downloaded PDF contains no trace of the marked content. 5. Verify the result — Open the downloaded PDF. Press Ctrl+A (select all) then Ctrl+C (copy) and paste into a text editor. The redacted content should not appear. If it does, the tool did not perform true redaction.

What content to redact — and what is often missed

Obvious content to redact: names, addresses, account numbers, social security numbers, medical information, and any text the document explicitly requires to be removed before release. Less obvious content that is frequently missed: Images containing text — If a page contains a photograph or scanned image that includes sensitive information, the image itself must be redacted (the pixel region overwritten), not just a text overlay above it. PDFs with embedded images store the original image data regardless of any text layer on top. Metadata referencing redacted content — PDF metadata (XMP and Document Information Dictionary) may contain the title, author, or keywords that include sensitive information. A document titled "Re: John Smith Drug Test Results" reveals a name in its metadata even if the body text has been redacted. Bookmarks and annotations — Bookmarks that quote redacted text, sticky note annotations containing the sensitive content, and form field values that reference it may all remain in the file after content-stream redaction. OCR text layers on scanned documents — A scanned PDF with an OCR text layer has two representations of its content: the image pixels and the invisible text stream. Redacting the image pixels without also removing the corresponding OCR text layer entries leaves the text extractable. Embedded attachments — PDFs can contain embedded files (via PDF/A-3 attachment mechanisms or general /EmbeddedFile objects). If these attachments contain the sensitive information, redacting the parent PDF body leaves the data present in the attachments.

Verifying that redaction worked

After redacting, always verify before distributing. These tests catch the most common failure modes: Text copy test — Press Ctrl+A to select all, Ctrl+C to copy, paste into a plain text editor. If the sensitive text appears in the paste output, the redaction was not applied to the content stream — only a visual overlay was placed. Search test — Open the PDF in a viewer with a search function (Ctrl+F). Search for a phrase from the redacted content. It should produce no results. PDF text extraction — Use a command-line tool like pdftotext or Apache PDFBox to extract all text: pdftotext document.pdf -. Review the output for the sensitive content. Metadata check — Use ExifTool: exiftool document.pdf to check that metadata fields do not contain the redacted content. Object inspection (advanced) — Tools like QPDF (qpdf --qdf document.pdf output.pdf) produce a human-readable version of the PDF object graph. Inspect the page content streams to confirm the sensitive operators are absent.

Correct redaction is a two-step process — mark and apply — with verification afterward. The mark step identifies what to remove; the apply step actually removes it from the content stream. Anything that skips the apply step is a whiteout, not a redaction. For sensitive documents, test the result with copy-paste, search, and a text extraction tool before distribution.

Frequently asked questions

What is the difference between redaction and adding a black box?

A black box is a visual overlay — it covers content on screen but leaves the underlying data in the file. True redaction permanently removes the data from the PDF's content stream. After correct redaction, the content cannot be recovered by removing a layer, copying text, or analysing the file bytes. After a black-box overlay, it can.

How do I know if my redaction actually worked?

After redacting, press Ctrl+A then Ctrl+C in any PDF viewer and paste into a text editor. If the "redacted" text appears, only a visual overlay was applied — the data is still present. For rigorous verification, use pdftotext to extract all text and search the output, and use ExifTool to check metadata.

Can I redact a scanned PDF?

Yes, but with caveats. A scanned PDF stores content as image pixels. Redacting a scanned PDF requires overwriting the image pixels in the sensitive area — not just placing an overlay. If the scanned PDF has had OCR applied, the invisible text layer must also be redacted separately. A proper redaction tool handles both; a simple drawing tool handles neither.

Does redacting a PDF change its file size?

Typically it reduces the file size slightly — the content data that was removed from the content stream is replaced by a simpler mark (a filled rectangle). For PDFs with large amounts of redacted image data (photos, scans), the reduction can be significant. For text-heavy PDFs, the difference is minimal.

Is online PDF redaction safe?

PDFTheGuide's redaction tool processes everything locally in your browser — your file never leaves your device. This is the safest option for sensitive documents. Cloud-based services upload your file to a server, which creates a custody chain and a potential exposure point. For classified or legally privileged documents, local processing (browser-based or desktop tools) is the correct choice.

Try it yourself

Everything in this article can be done directly in the free PDFTheGuide editor — no account required.

Redact PDF Whiteout PDF content Annotate PDF Open editor