SANITIZE

CertifiedNIST SP 800-88 Rev. 2 · DoD 5220.22-M · IEEE 2883

More Passes Do Not Mean More Security

The three-pass and thirty-five-pass methods still sold as premium options solve a problem that stopped existing with the drives they were designed for.

Ask for an erasure quote and you will usually be offered a choice framed as a strength ladder: one pass, three passes, seven, sometimes thirty-five. The implication is that more passes are safer and that the customer is choosing how much risk to accept. On any drive manufactured this century, that ladder is fiction.

Where the idea came from

Multi-pass methods were designed for magnetic media of the 1980s and 1990s, where track density was low enough that a previous write could in principle leave a readable magnetic remnant beside the new one. The patterns in the Gutmann method are not arbitrary; each targets a specific encoding scheme used by drives of that era. Modern perpendicular recording does not behave that way, and the encodings those patterns targeted have not shipped in decades.

What passes cost

Time, in direct proportion. A three-pass wipe of a 4 TB drive takes three times as long as a single pass, and a floor of machines running seven passes is a week of bench time spent on a threat model that no longer applies. That time is the real reason to care: an organisation that cannot finish its disposal backlog leaves un-erased assets sitting in a cage, which is a far larger exposure than the theoretical remnant the extra passes were meant to address.

What actually improves assurance

  • Verification. Reading the drive back and comparing what is there against what was written is the only step that turns "we ran a wipe" into "the wipe worked".
  • The right method for the media. On flash, passes are close to meaningless because the controller decides which physical cells a write touches — the drive's own sanitize command is the answer.
  • Covering the whole addressable space, including reallocated sectors and host-protected areas that a naive overwrite skips.
  • Recording the result per device, so a failure is visible rather than averaged into a batch.
One verified pass is stronger evidence than seven unverified ones.

The multi-pass methods are still in the software's register, because some policies and some regulators name them explicitly and a tool that cannot satisfy the paperwork is no use. But the register says what each one does, and none of them is described as safer than a verified single pass on media where that is sufficient.

Tell us what you need

Looking For Something Specific

Ask directly. A question about a standard usually gets a better answer than a page can give.

See Sample Certificates

Goes straight to our engineers. No newsletter, no call centre.