This article presents a discussion of research that has evaluated the effectiveness of batterer programs and attempts to show why the research has had difficulty reaching a consensus about the effectiveness of the programs. The discussion begins with an illustration of the conceptual and methodological challenges in evaluating the effectiveness of batterer programs and reasons for some of the contradictory outcomes. The review found that evaluation is not an objective or purely scientific process that produces unbiased and conclusive results, but rather is a process with a subjective outcome. This subjectivity of the evaluation approaches makes it hard to generalize results and apply them to policy. A review of over 40 published program evaluations that have attempted to address the effectiveness of batterer programs found that the evaluations are compromised by selection bias, low response rates, short follow-up periods, no or weak control groups, and no calculations of effect size. In order to address some of the conceptual issues and methodological shortcomings of previous evaluations, the authors developed a multisite evaluation of programs in 4 cities that consisted of a 4-year follow-up, starting at program intake, with periodic interviews of 840 men and their female partners. Results of the evaluation show that there is a clear de-escalation of reassault and other abuse, that the vast majority of men do reach sustained nonviolence, and that about 20 percent of batterers continuously reassault. The evaluation also suggests that the predominant gender-based cognitive-behavioral approach to counseling may be appropriate for the majority of men. The research found that program effectiveness depends on the intervention system used of which the batterer program is a part. Program and evaluation implications are discussed. References