The experiment involved police approaching domestic assaults in one of three ways: mediation, arrest, and requiring the assailant to leave the premises. The study found that the assailants arrested were only half as likely as assailants handled by other methods to repeat their assaults. Recidivism was measured by subsequent interviews with victims and an examination of arrest records. The Minneapolis Police Department now has a policy of arresting alleged assailants if there are visible signs of injury to the victims. Portrayed interviews with some Minneapolis officers indicate their unhappiness with the policy and their desire for greater discretion in their approach to particular incidents. The panel interviewed consists of Lawrence Sherman, University of Maryland, who conducted the experimental study; George Napper, Police Commissioner of Atlanta, Ga.; and Barbara Hart, National Coalition Against Domestic Violence. Napper argues for an emphasis on treatment, with police acting as trained referral agents. He challenges the findings of the Minneapolis study, based on the fact that the mediation or treatment approach was ill-defined and probably consisted of little more than some superficial police advice to the assailant. Hart argues for a combination of arrest and treatment, since effective treatment is long term, and court action is required to enforce treatment participation. Hart cites a program in Duluth, Minn., which combines arrests and criminal justice processing with treatment. Participants have not reoffended for 1 year after treatment. Napper notes that the Atlanta police do make arrests when the domestic assault is an apparent felony, i.e., a weapon or severe injury are involved. The panel agrees that further empirical evidence is needed to indicate the effects of particular police policies in dealing with domestic assault.