Victim interviews sought to determine the accuracy of police reports, especially with regard to police officer statements about the victim's preferred police action, and to obtain the victim's perspective about the dynamics of the domestic violence incident, the police-citizen encounter, and the recurrence of violence. Data analysis revealed a somewhat complex pattern of factors that correlated with decisions to arrest or not arrest offenders. The three primary determinants appeared to be the presence of others at the scene, whether the victim lived with the offender, and the victim's preference. The presence of bystanders during abuse dramatically increased the chances that an arrest would be made. Arrests were more than twice as likely when the offender and the victim shared the same residence. About 34 percent of the victims wanted to prosecute abuse perpetrators. When the victim's preference was to do nothing or merely talk or be advised of her rights, arrests were made in only 21 percent of the cases. The nature of injury was significantly related to the probability of arrest. Victim interviews demonstrated a high level of satisfaction with police department responses. 20 references and 8 tables