Compared to the North Carolina Assessment of Risk (NCAR), whose predictive validity has been found to vary by both gender and race/ethnicity, the JRM had higher levels of predictive validity with all offenders. This improvement was due to the expansion of measured risk and the development of the multiple assessment strategy. The statistically significant, albeit modest, relationship between the dynamic factors and static risk scales shows that they measure overlapping, yet distinct, risk constructs. This suggests that both contribute to a valid assessment of the risk of reoffending. Gender-based differences in risk assessment were explained by gendered patterns of referral to out-of-home placements. The findings suggest that risk assessment can be improved by including measures related to the behavior and demeanor of offenders as well as the cooperation of their parents/caretakers. The predictive validity of the NCAR and the JRM were assessed on a statewide proportional sample of delinquent juveniles from North Carolina. Initial data collection began during February 2002. Follow-up data collection began during February 2003. Sixty-eight percent of the sample was male, and the sample was evenly balanced between African-American and non-Latino White youths, with the remaining youths classified as other racial and ethnic groups. The JRM includes 3 static factors (first offense prior to age 12, number of prior referrals, and history of running away) and 11 dynamic factors, which include such measures as drug/alcohol abuse, peer delinquency, parental supervision, impulsivity, and pattern of hostile/aggressive behavior. 5 tables and 40 references