\”Without prejudice\”: the two words that do not protect you the way you think

A manager forwards a difficult email to a colleague. She is nervous about the wording, so she types “Without prejudice” at the top and hits send. Eighteen months later that same email is sitting in a bundle in front of a tribunal, quoted in full, with the label doing precisely nothing. The two words she trusted were decoration.

This happens constantly, and it happens because the phrase has drifted from a narrow legal rule into an office reflex. People attach “Without prejudice” to grievance replies, performance emails, commercial threats, and internal chats, on the theory that the label buys them a shield. It does not. The label is not a spell. It describes a category of communication, and if your message does not fall inside that category, writing the words changes nothing about whether a court can read them.

What the rule actually protects

The without-prejudice rule is a privilege that keeps genuine settlement communications out of evidence. The policy behind it is simple and sensible: people should be able to try to settle a dispute without every concession they float being weaponised later if the talks fail. So a real offer to compromise, made to resolve an existing dispute, is generally protected whether or not you label it, and the courts look at substance over the sticker on top.

Two conditions have to be met, and both matter:

• There is an existing dispute. Not a vague future disagreement, and not a routine negotiation over a deal that is going fine. There has to be something the parties are actually in contention about, with litigation or a formal complaint realistically in view.
• The communication is a genuine attempt to settle it. The email has to be part of trying to resolve that dispute, containing concessions or offers made in that spirit.

Miss either condition and the label is inert. A stern email to a supplier who has not yet raised any dispute is ordinary correspondence, admissible in full, no matter how many times you type the magic words. An internal message between two colleagues is not a settlement communication with anyone, so the label is meaningless there by definition.

The exception that surprises people: unambiguous impropriety

Even when a communication genuinely is a protected settlement discussion, the shield can fall away. Courts recognise an exception for what is often called “unambiguous impropriety.” If a party uses the cover of settlement talks to make an overt threat, to blackmail, or to tell a clear untruth, the protection does not extend to that conduct. The privilege exists to encourage honest compromise, and it will not be stretched to launder something improper said under its cover. The bar is high and courts apply it cautiously, but it exists, and it is the reason “we said it in a without-prejudice call” is not a safe place to say something you would not want read aloud.

Admissions do not become invisible

Here is the part that costs organisations the most. A statement of fact, especially a damaging admission, tends to stay admissible when it lives outside a real settlement context. The label cannot retroactively convert an admission into a protected offer.

Consider the aircraft manufacturer whose internal messages became central to a USD 2.5 billion deferred prosecution agreement. Among the lines that surfaced was an employee writing that he “basically lied to the regulators.” No label on earth reclassifies that sentence as a settlement offer. It is an admission, made internally, about past conduct, and it reads as one to anyone who finds it. The same logic reaches the ordinary emails your team sends every week. If someone writes down that a process was ignored, that a customer was misled, or that a known defect shipped anyway, the sentence is evidence of the fact it describes. Typing “Without prejudice” above it adds a false sense of safety and nothing else.

This is general information rather than legal advice, and the precise boundaries vary by jurisdiction and forum. The practical point holds across all of them: the protection follows the nature of the communication, and the label follows nothing.

The usable artifact: a four-question test

Before anyone types “Without prejudice,” run the message through these four questions. If the honest answer to any of the first three is no, the label does not apply and should come off.

What to write instead when the label does not apply

When the four-question test says the label is wrong, the fix is rarely a better label. It is cleaner content. A few practical substitutions:

• Drop the “Without prejudice” header on any email that is not a settlement offer. Keeping it signals to a future reader that you thought the message was risky, which is worse than saying nothing.
• Where you are genuinely negotiating a live dispute, use “Without prejudice save as to costs” only when you mean it, and keep the offer factual.
• Replace speculation and admissions of fault with what you actually know. “We clearly messed this up and probably breached the contract” is an admission. “Here is what happened and the steps we are taking” is a record you can stand behind.

The point of the test is to move the decision earlier, to the moment of writing, where it is cheap, rather than to the moment of disclosure, where it is not.

Where a pre-send check fits

Most of these emails are not written by bad actors. They are written by well-intentioned people under time pressure who reach for a label they half-remember and a phrasing they have not thought through. That is exactly the gap a pre-send check is built for. VerbaPulse reads a message the instant before it leaves and flags the careless line the sender did not notice: the misapplied “Without prejudice,” the throwaway admission, the sentence that reads as a threat.

A short, phrase-level example of what that looks like in practice: on an internal email carrying a “Without prejudice” header, the flag is on the label itself, with a suggestion to remove it because there is no settlement context here. Nothing elaborate, just the specific words that will not do what the writer assumed.

It is worth being precise about scope. A pre-send check is a front-end shield against accidental human risk. It is not an adversarial security boundary, and it will not stop someone determined to send something damaging. It complements the archiving and supervision systems (the Smarsh and Proofpoint layer) by catching issues before they ever reach those queues, so fewer of them do. We wrote more about the legal-team angle on our page on how “without prejudice” actually works.

The takeaway

Stop treating “Without prejudice” as insurance. Run the four-question test before you type it, take the label off anything that is not a genuine settlement communication, and write the underlying sentence as if it will be read aloud with no privilege at all, because quite often it will be. The label protects a narrow category of honest settlement talk. Everything else is just an email, and your people should write it that way.