The developmental taxonomy proposed by Moffitt (1993, 2006) holds that there are two main trajectories of offending: a life-course-persistent (LCP) and an adolescent-limited (AL) pattern. A bulk of research, primarily for males, supports the LCP and AL conceptualization. Childhood risk factors behind the development of LCP and AL offending seem to be quite similar for males and females. If the model proposed by Moffitt is correct, however, possibilities to predict future antisocial behavior should differ between the sexes. Early childhood problems which predict future LCP are more common among males. Hence it should be possible early to forecast LCP trajectories for males, but not for females. Rather, what happens from childhood to early adolescence should be predictive of females' future antisocial behavior. We tested the possibility to predict middle adolescent normbreaking and adult criminality from late childhood and late childhood to early adolescence problem indicators, respectively, for the 1,000 males and females in the longitudinal research program Individual Development and Adjustment (Magnusson et al., 1973). Analyses showed that male offending could be better and earlier predicted than female offending, but what happens in early adolescence mattered for later normbreaking and criminality for both females and males.