The analysis was restricted to those officers who had trained at the Academy since the introduction of the PROVE program in January 1994. A total of 437 recruit records were made available for analysis. A standard multiple regression was performed, with Academy performance as the dependent variable and the composite psychometric scores, education/employment scores, and panel interview scores as independent variables. Psychometric test results and education/employment rating were found to be the best predictors of performance at the Academy; panel interview rating contributed little to predicting Academy performance. Psychometric test results and education did not correlate, indicating they are independent measures of performance. Some of the psychometric tests that formed the composite score for each applicant had poor predictive value for Academy performance. The 16PF (personality test) did not assist in screening applicants in or out of the selection process. No consistent test profile could be found for poor performers at the Academy. The study concludes that the current selection criteria and selection process used by the Queensland Police Service should be reviewed. Particular attention should be given to the interview process and the use of personality tests. Further, there should be a revision of the weighting of the criteria (currently 33 percent each) in the final composite score for each applicant. In revising the current selection criteria, consideration should also be given to the sequence of "hurdles" used in the selection process. The most cost-efficient assessment with good predictive power should be used as early in the screening process as possible and the more expensive assessments in the final stages of selection. 4 tables, 3 references, and appended education/employment assessment for January and May 1995