In this paper, we describe a quantitative summary of 12 twin (n=3795 twin pairs and 3 adoption studies=338 adoptees) published since 1975 which provided 21 estimates of the heritability of antisocial behavior. Medium to large effect sizes were found for genetic influences across studies, with approximately 50% of the variance in measures of antisocial behavior attributable to genetic effects. Although effect sizes did not vary across different definitions of antisocial behavior (criminality, aggression, or antisocial personality), significantly larger estimates of genetic effect were found for severe manifestations of antisocial behavior. The importance of severity was further underscored by the significantly larger effects obtained in studies using clinic-referred samples compared to the effects obtained in studies using volunteer samples. Demographic characteristics of the samples did not influence effect sizes, although studies using more stringent methodology tended to find larger effects. These results must be interpreted in light of the small literature that was suitable for the meta-analysis due to numerous methodological limitations in existing studies.