Evaluative and predictive judgments are based on diagnostic attributions regarding the causes of criminal behavior. Two experiments using realistic case materials and expert decision makers from two State parole boards are reported that explore attributions, evaluations, and predictions in parole decisions as a function of case evidence, structural position of the parole decision within the criminal justice system, and guidelines recommendations. Results showed good support for the attributional model, in that evaluations for punishment are important only when judges give no minimum sentence. However, reactions of a drug habit were difficult to interpret. Guidelines seemed to have less impact than case based recommendations. 3 tables and 18 references. (Publisher abstract)