A review of the research literature concerning the effectiveness of treatment or prevention programs for domestic violence, conducted by the National Academy of Sciences, revealed 2,000 published studies. Of these, only seven studies were conducted with enough methodological rigor to warrant consideration by the Academy. These seven studies tested the deterrent effectiveness of mandatory arrest policies for domestic violence cases; they are reviewed here. The first of the seven studies was the Minneapolis domestic violence experiment, which is among the most visible and highly cited research in criminology. This study was lauded for its methodological rigor and the findings indicated that suspects who were arrested for misdemeanor spouse assault were 50 percent less likely to re-offend than suspects who were not arrested. Six other studies attempted to replicate these findings in other cities, with similar methodological rigor. Their results, when taken together, indicate that arrest in all misdemeanor cases of spouse assault will not produce a discernable effect on recidivism. However, two recent studies that employ different approaches for systematically combining results across multiple sites have found significant deterrent effects from arrest. These two studies show that methodological rigor is important throughout the scientific process, not just during data collection and analysis. The authors conclude that there will be no new studies of the deterrent effect of mandatory arrest policies because such studies took advantage of a particular historical moment and because policy debate has shifted from arrest policies to what should be done with suspects and victims after an arrest. Unfortunately, contemporary policy debate concerning appropriate responses to domestic violence does not appear to be derived from, or to reply upon, systematic empirical research in the standards of the National Academy of Sciences. Notes, references