Over 40 published program evaluations have attempted to address the effectiveness of "batterer programs" in preventing reassaults. Summaries and meta-analysis of these evaluations suggest little or no "program effect." Methodological shortcomings, however, compromise most of these quasi-experimental evaluations. Three recent experimental studies appear to confirm little or no effect, but implementation problems, intention-to-treat design, and sample attrition limit these results. A longitudinal 4-year follow-up evaluation in four cities poses additional considerations and evidence of at least a moderate program effect. There is a clear deescalation of reassault and other abuse, the vast majority of men do reach sustained nonviolence, and about 20% continuously reassault. The prevailing cognitive-behavioral approach appears appropriate for most of the men, but the following enhancements are warranted: swift and certain court response for violations, intensive programming for high-risk men, and ongoing monitoring of risk. Program effectiveness depends substantially on the intervention system of which the program is a part.