The authors examine 1980 to 2003 trends in female-to-male interpersonal violence reported in Uniform Crime Reports (UCR) arrest statistics and National Crime Victimization Survey (NCVS) victimization data. Augmented Dickey-Fuller time-series techniques and intuitive plot displays show much overlap yet differences in each source's portrayal of trends in female violence levels and the gender gap. Both sources show little or no change in the gender gap for homicide and rape/sexual assault, whereas UCR police counts show a sharp rise in female-to-male arrests for criminal assault during the past one to two decades—but that rise is not borne out in NCVS counts. Net-widening policy shifts have apparently escalated the arrest proneness of females for "criminal assault" (e.g., policing physical attacks/threats of marginal seriousness that women in relative terms are more likely to commit); rather than women having become any more violent, official data increasingly mask differences in violent offending by men and women.