The study found that arrested perpetrators of assaults on their wives were more likely to reoffend, but this effect was attributable to prearrest differences in perpetrators' risk of reoffending. There was no evidence that arrest for wife assault increased reoffending at any level of risk. The findings suggest that officers tended to arrest higher risk cases based on the seriousness of the offense as measured by victim injury. The study did not measure whether offenders who were not arrested but might have had a high risk for reoffending might have reoffended. Despite the fact that the jurisdiction studied had an official presumptive arrest policy for domestic violence, police officers arrested only approximately half of the perpetrators. Survival analyses indicated that arrest might have delayed reoffending among relatively low-risk cases. This delay effect might be expanded by providing police with an actuarial tool for a more accurate field assessment of an offender's risk for reoffending. As part of a larger study of wife-assault reoffending, this study coded information on 589 men identified in police records as perpetrators of a physical assault on a current or former wife or common-law wife, as well as a credible threat of death against such a victim with a weapon in her presence. The incident closest to the end of 1996 was selected as the index assault. Researchers coded all information in the police reports, including items contained in the ODARA, police response, characteristics of the perpetrator and victim as well as their relationship, and variables previously linked with arrest decisions. 1 table, 1 figure, and 52 references