Physical assaults against women result in more than 5,000 deaths and 1 million nonfatal injuries per year in the United States. Data from the National Crime Victimization Survey, 1992-1995, were used to test the association between injury risk and self-protective behaviors, while controlling for victim, offender, and crime-related characteristics. Unlike in prior studies, a self-protective behavior measure that accounted for the temporal sequencing of the occurrence of injuries and self-protective behaviors was used. The study also examined whether the effect of self-protective behaviors varied as a function of victim-offender relationship status. The sample included 3,206 incidents in which females were physically assaulted by a lone male offender within the previous 6 months. Multivariate results revealed that women who used self-protective measures were less likely to be injured than were women who did not use self-protective measures or who did so only after being injured. The effect of self-protective behaviors on risk of injury did not vary as a function of the victim-offender relationship. The inverse association found between self-protective behaviors and injury risk differs from those of previous studies. Owing to inconsistent findings across studies, caution should be used when making recommendations to women regarding whether or not they should use self-protective behaviors during a physical assault.