The 137 participants were recruited from a shelter for battered women in a medium-sized midwestern city. They were interviewed several times over 2 years. Their police and court records were also examined. In the 2-year period after leaving the shelter, 42 percent of the women indicated that the police had been contacted at least once because of the original assailant's violence. Results revealed that a low rate of agreement among the sources regarding arrest and conviction. The victims were more likely than the other sources to indicate that an arrest or a conviction occurred. The overall rate of agreement between victims and official records regarding whether the assailant was arrested during a given time period was 57 percent, largely because both sources agreed that arrest did not occur. In only 5 percent of the cases did the victim and official records agree that an arrest had taken place and on the time period of that arrest. Overall, agreement about conviction in a given time period was 80 percent, mainly because conviction was a rare event. Findings indicated the need for further research to understand more accurately the pattern of disagreement between official and victim reports. In the meantime, researchers and policy makers should be cautious in reaching conclusions based only on official or victim reports of the criminal justice system response to domestic assault. Tables, notes, and 14 references (Author abstract modified)