This study uses data on individual arrests for domestic violence from the North Carolina Administrative Office of the Courts to assess effects of prosecution, conviction, jail sentences and fines on the probability of a repeat arrest and conviction within two yeas of the index arrest. After accounting for endogeneity of penalties, we find some evidence that penalizing persons convicted of a domestic violence offense deters future offense, but no deterrence effect for the much larger number of persons who are arrested for domestic violence but whose offenses are not prosecuted or if prosecuted do not result in a conviction.