Objective: Antisocial behavior that begins in mid- to late adolescence does not fit into commonly accepted taxonomies of antisocial behavior, yet it clearly exists. This study examined how this course of antisocial behavior compares with persisting (beginning by early adolescence and continuing through late adolescence) and desisting (stopping by mid-adolescence) antisocial behavior in terms of risk for later substance dependence and background risk factors (gender, IQ, socioeconomic status, parental antisocial behavior, and parental divorce). Method: A population-based sample of twins from the Minnesota Twin Family Study, evaluated at ages 17 and 20, was used. Results: The results indicated that youths with late onsets were similar to those with persisting antisocial behavior and that both groups were at higher risk of later nicotine, alcohol, and cannabis dependence than controls; both also had similarly high levels of background risk factors. The late-onset group included a significant over representation of females, whereas the persisting and desisting groups included more males. Conclusions: Late-onset antisocial behavior has many of the same negative correlates of persisting antisocial behavior but includes significantly more females. Although they are excluded from the diagnosis of antisocial personality disorder, these youths have clinically significant problems similar to those with this diagnosis.