Krueger and colleagues (2007) have elaborated on a hierarchical structural model of externalizing psychopathology, with a general overarching factor and 2 residual liability factors reflecting callous-aggression and substance misuse. The current study used an independent set of symptom and behavior indicators (items derived from the Minnesota Multiphasic Personality Inventory–2-Restructured Form) to test alternative models of externalizing. In mixed-sex samples of correctional inmates (n = 42,290), forensic criminal defendants (n = 1,065), and community participants (n = 2,214), a bifactor model based on Krueger, Markon, Patrick, Benning, and Kramer (2007) was associated with better model fit relative to a 1-factor model, 2-factor model, and a bifactor model with 2 random bifactors. Moreover, the Krueger et al. model was largely invariant across sex, race, and time. Finally, external validity analyses from the forensic sample revealed that the latent factors were associated with a range of forensically relevant outcomes. More specifically, those high on general externalizing were less likely to be deemed mentally ill, incompetent, or insane in court-ordered evaluations. Those high on the callous-aggression bifactor were more likely to have violent crime convictions. Those high on the residual substance misuse factor were more likely to be recommended for drug intervention in-lieu of conviction and be clinically diagnosed with drug dependence. In conclusion, the externalizing spectrum appears robust regardless of settings and indicators examined, and its domains are associated with theoretically expected outcomes.