First, three methods of scoring a delinquency self-report measure (frequency, variety, and seriousness) were compared to assess their differential relevance to the prediction of delinquency proneness. Second, a multivariate model was examined to assess its explanatory utility for identification of delinquency proneness. Findings, replicated through a series of regression analyses, demonstrate that age of onset is an important risk factor for involvement in delinquent behaviors. The only other significant predictive variables were those related to family system characteristics. Psychosocial indicators added little to the predictive model. A specific factor model accounted for more total variance than did the risk-count model. The advantage of self-reported delinquent behavior over official statistics is discussed, as is the comparability of self-reported scoring procedures. How self-reported delinquency is scored does not appear to be as critical as previously thought. 2 tables and 33 references. (Author abstract modified)