Different types of evaluations-–non-experimental, quasi-experimental, and experimental evaluations--are described and the results of four previous experimental evaluations of batterer intervention programs are enumerated. Results of these evaluations indicated that two of the programs evidenced no reductions in batterer recidivism for participants while the other two programs only showed modest reductions in recidivism rates among participants. Evaluations of different program types and structures further indicated no differences in effectiveness rates for different types of batterer intervention models. The authors conclude that efforts should continue on experimentation and program development for batterer intervention programs, as there does appear to be the promise of modest improvements for well-designed programs. Three main research questions shaped the analysis: (1) are batterers held accountable for their crime; (2) are victims safe; and (3) has the batterer changed his attitudes and behavior? It is important to evaluate batterer programs because, among other reasons, one of the strongest predictors that a battered woman will remain with her batterer is that he has sought counseling. As such, it is crucial that the counseling actually works to reduce violent behavior in the home. The challenges that emerge when researching batterer intervention programs are discussed, including the problems of how to measure victim safety and how to address co-occurring issues, such as substance abuse, unemployment, and mental disorders that may be aggravating the abuse. References, appendix