AN INITIAL REVIEW OF THE LITERATURE ON CLINICAL JUDGMENT FINDS THAT PAROLE DECISIONMAKERS ARE PORTRAYED AS SUBJECTIVE AND UNWILLING TO PROFIT FROM SCIENTIFIC EVIDENCE. IN STUDY 1, BASE-RATE DATA ON A GROUP OF 100 OFFENDERS WAS GIVEN TO 112 CRIMINOLOGY STUDENTS. THE STUDENTS MADE NO DISTINCTION IN RISK EVALUATIONS BETWEEN PAROLEES FROM A GROUP WITH A KNOWN RECIDIVISM HIGH RATE AND THOSE FROM A GROUP WITH A LOW RATE. IN STUDIES 2 AND 3, HOWEVER, BOTH EXPERTS AND STUDENTS USED RISK INFORMATION PROVIDED WITH THE INDIVIDUAL CASE. IN STUDY 2, A TOTAL OF 210 CASE SUMMARIES OF PAROLE APPLICANTS WERE REVIEWED. OF THESE, 62 PERCENT CONTAINED INFORMAL ASSESSMENTS OF PAROLE RISK. ALL WERE USED BY THE HEARING EXAMINER. STUDY 3 DREW 12 CASES FROM THE TOTAL SAMPLE OF 210 AND GAVE THEM TO 75 CRIMINOLOGY STUDENTS. HOWEVER, RISK STATEMENTS WERE EITHER PRESENTED VERBALLY OR IN A STATISTICAL MANNER. THE STUDENTS USED THE INFORMATION REGARDLESS OF THE FORM, BUT WERE NOT AS SENSITIVE TO IT AS THE EXPERTS. THE STUDIES ALSO SHOWED THAT NEGATIVE STATEMENTS CARRIED MORE WEIGHT THAN POSITIVE STATEMENTS AND THAT VERBALLY-STATED RISK PREDICTIONS SEEM MORE INFLUENTIAL THAN EITHER NUMERICALLY-STATED OR CLINICALLY-DERIVED RISK STATEMENTS. STUDY STATISTICS ARE DISCUSSED IN DETAIL. AN EXTENSIVE LIST OF REFERENCES IS APPENDED.