This research, inspired by the cognitive interviewing literature, investigates misreporting of intimate partner violence when survey participants interpret items in unintended ways. In over 23 per cent of victimizations reported by university-aged males and in over 12 per cent of victimizations reported by females, follow-up questions revealed that purported violence was either accidental or done in a manner where neither partner took the event seriously. The problem was worse for perpetration reports where over 47 per cent of male reports and over 17 per cent of female reports were endorsed in a manner unintended by instrument design. The magnitude of the problem, together with its gendered nature, suggests that misreporting of this type is a substantial problem having the potential to negatively affect the testing of partner-violence theories.