The research considered some of the implications of two positions regarding criminal propensity. One position is that the causes of variation in offending behavior can be traced to variation in one or more causal traits. Other theorists contend that more than one type of offender exists and that more than one causal mechanism operates to explain offending behavior. The analysis considered the arrest records of all the youths as of November 1994, which represented an average follow-up period of just under 6 years. Persistence in crime was defined as the length of the interval between the release from the training school and the first post-release arrest; those with the shortest intervals exhibited a higher propensity to persist. Results revealed support for both positions regarding criminal propensity, in that it indicated some similarities and some differences in the correlates of persistence. However, the differences were evident only when the threshold for late first adjudication was set to age 12. When this threshold was raised to higher ages, the differences disappeared. Results also revealed that child abuse was criminogenic for the group that was first adjudicated early, but the effect of child abuse was trivial in the much larger late first-adjudication group. Findings suggested that a significant amount of crime could be prevented by intervening in families in which poor parenting or abusive parenting is likely to occur. Figures, tables, footnotes, and 54 references (Author abstract modified)