The Adolescent Risk-taking Questionnaire (ARQ) was developed to comprehensively assess adolescent risk-taking beliefs and behaviours. Research has shown that it is a reliable instrument with strong construct validity. The current study investigated the convergent and discriminant validity of the ARQ by comparing responses of a sample of 52 male adolescents from a juvenile justice centre and 211 school-attending male adolescents. Consistent with past research, the juvenile justice centre youth reported significantly higher levels of depression and less optimal levels of parent attachment compared to school-attending youth. Convergent validity of the ARQ was demonstrated through significant relationships between attachment and risk-taking. Specifically, those adolescents with less optimal parent attachment were more likely to engage in risk-taking behaviour and were less likely to perceive the behaviours included in the ARQ as risky compared to those with more optimal levels of parent attachment. Discriminant validity of the ARQ was demonstrated through significant differences between the two groups on risk beliefs and behaviours. Responses on the ARQ indicated that youth from the juvenile justice centre were significantly more likely to engage in risky behaviours, if given the opportunity, than school-based youth. They also perceived the behaviours to be less risky. It was concluded that the ARQ has good convergent and discriminant validity.