Mandatory and preferred arrest policies for domestic violence (DV) have led to an increase in DV arrests and prosecutions of male and female suspects. The strain of this increase on criminal justice resources suggests a need for prioritizing DV cases and determining suspects most at risk for recidivism and victimization. The present study sought to address a largely ignored question in the field of DV: who is more likely to recidivate, male or female DV offenders involved with the criminal justice system? Two forms of recidivism were coded using subsequent police reports for DV: recidivism as a suspect and recidivism as a victim. The results indicate that there are significant differences in the ways in which male and female DV offenders recidivate. Males were more likely to recidivate as a suspect and females were more likely to be listed as a future DV victim in police reports. Male and female offenders also significantly differed in the frequency of their recidivism. Gender, independent of other factors, was reliably associated with both forms of recidivism and a similar pattern of recidivism was found in two different cities. There was, however, a small group of female offenders in both cities who appear to be primary aggressors. Another grouping of female offenders appears to be involved in relationships characterized by mutual aggression and bi-directional violence. Together these two groups account for roughly a quarter of the women in our sample. The implications of the findings for arrest practices, treatment and interventions, and using gender as a risk factor are discussed. Validity and reliability problems with using police reports as a measure of DV recidivism are also reviewed. (Contains 3 tables and 4 notes.)