Externalizing traits or behaviors are typically assessed by self-report scales or criminal records. Few genetically informative studies have used both methods to determine whether they assess the same genetic or environmental risk factors.

We examined 442 male Swedish twin pairs with self-reported externalizing behaviors at age 16–17 years [externalizing traits (EXT), self-reported delinquency (SRD), impulsivity (IMP), grandiosity (GRD) and callousness (CLS)] and criminal behavior (CB) from the National Suspect Registry from age 13 to 25 years. Multivariate structural equation modeling was conducted with Mx.

The best-fit model contained one genetic, one shared environmental and two non-shared environmental common factors, and variable specific genetic and non-shared environmental factors. The risk for CB was influenced substantially by both genetic (a2 = 0.48) and familial–environmental factors (c2 = 0.22). About one-third of the genetic risk for CB but all of the shared environmental risk was indexed by the self-report measures. The degree to which the individual measures reflected genetic versus familial–environmental risks for CB varied widely. GRD and CLS were correlated with CB mainly through common genetic risk factors. SRD and CB covaried largely because of shared familial–environmental factors. For EXT and IMP, observed correlations with CB resulted in about equal parts from shared genetic and shared familial–environmental factors.

In adolescence, measures of grandiose and callous temperament best tap the genetic liability to CB. Measures of antisocial behaviors better index familial–environmental risks for CB. A substantial proportion of the genetic risk to CB was not well reflected in any of the self-report measures.