Using an existing data base covering variables collected on 4,146 California Youth Authority wards, the project (1) derived seven information dimensions of variable categories believed relevant to the clinical context (individual case history, offense-specific, academic, intelligence, vocational, social/psychological, and psychological areas); (2) compared four prediction strategies (multiple regression, predictive attribute analysis, association analysis, and the Burgess method) in terms of their appropriateness to the clinical format of the DPPI as well as their predictive efficiency; (3) designed three alternate formats for presentation of dimensional data; and (4) field-tested these three formats to determine their relative utility and appropriateness for decisionmaking. Comparison of multiple regression and the Burgess technique show low predictive power even from the best dimension, and substantial variation in the power of predictions developed from the seven dimensions. Analysis of the predictive power of all four techniques found the predictive attribute analysis displays the highest predictive power on both construction and validation. The Burgess technique performed least efficiently of all techniques. A major feature of the analysis was the poor predictive power of the instruments used in the study. A survey of corrections practitioners found few with an interest in developing a predictive instrument. Further research into this area must first consider clinical decisions, processes, and needs. One figure, 16 tables, and 85 references are included.