A randomized trial was conducted to test the efficacy and sufficiency of parent-child interaction therapy (PCIT) in preventing re-reports of physical abuse among abusive parents. Physically abusive parents (N=110) were randomly assigned to one of three intervention conditions: (a) PCIT, (b) PCIT plus individualized enhanced services, or (c) a standard community-based parenting group. Participants had multiple past child welfare reports, severe parent-to-child violence, low household income, and significant levels of depression, substance abuse, and antisocial behavior. At a median follow-up of 850 days, 19% of parents assigned to PCIT had a re-report for physical abuse compared with 49% of parents assigned to the standard community group. Additional enhanced services did not improve the efficacy of PCIT. The relative superiority of PCIT was mediated by greater reduction in negative parent-child interactions, consistent with the PCIT change model.