Although there are several dozen evaluations of batterer treatment programs, few have used methodologies that are appropriate to addressing the issue of whether treatment is effective; however, among the handful of quasi-experiments and true experiments, there is fairly consistent evidence that treatment works and that the effect of treatment is substantial. Regarding the second question, there is little evidence to date that one form of treatment is superior to another or that longer programs turn out less violent graduates than shorter ones. Regarding the third and last question, there are bases for hypothesizing that some batterers may fare better in treatment (or fare better in certain types of treatment) than others; however, empirical verification has been limited to date. The authors recommend that several standards be applied to future investigations regarding whether treatment has an effect on violence. First, randomized experiments should be the design of choice. Second, measures and follow-up intervals should be standardized so that results can be compared across studies. Third, investigations of the effects of batterer treatment should be explicit in defining the standard against which treatment is being evaluated. Fourth, research should find ways to minimize attrition from treatment programs. Fifth, research ought to make explicit issues that may restrict the extent to which their findings can be generalized. Finally, researchers should find ways to maximize interview response rates when interviewing victims about continuing abusive behavior from their spouses. 6 tables and 59 references