Risk-need assessment is widely accepted as best practice with juvenile offenders and is underpinned by a healthy research literature on risk assessment inventories. Previous studies have found both similarities and differences on risk measures when gender and racial/ethnic subgroups have been compared. Differential validity has been examined, but differential prediction has been overlooked. The current study undertook gender and ethnic comparisons for a large sample (N = 3568) of community based juvenile offenders who were evaluated using the Australian Adaptation of the Youth Level of Service/Case Management Inventory (YLS/CM-AA). Analyses showed various gender and ethnic differences at the item level, across domain scores and on the total inventory score, but not for validity indices. However, 1 year re-offending rates for youth in three classification categories (low, moderate, high) varied by gender and ethnicity. The findings were related to contemporary understandings of the risk factors for offending and the dynamics of crime for gender/ethnic subgroups. It is argued that in spite of these subgroup differences, a generic inventory such as the YLS/CM-AA can be used fairly with various subgroups. Recommendations for how this could be accomplished are provided.