Data were obtained from 308 men and their partners collected at five, 3-month intervals. Time-varying situational and behavioral risk factors, as well as time-invariant individual characteristics, were examined. The most influential risk markers in terms of relative risk and level of statistical significance were time-varying: two measures of the man's drunkenness during the follow-up interval in which the reassault occurred. Other included time-varying batterer characteristics had no significant effect on reassault. Two significant time-invariant batterer risk factors were severe psychopathology and a history of non-domestic violence arrests, both measured at intake. Findings suggest that batterers' drinking behavior after program intake may provide an important and easily observed "marker" for risk of reassault. The prediction of reassault with individual risk factors at program intake remains problematical. 2 tables and 23 references