Validation and cross-validation studies were conducted using the Minnesota Multiphasic Personality Inventory (MMPI) and Inwald Personality Inventory (IPI) to predict job performance for 698 urban male police officers who completed a six-month training academy. Job performance criteria evaluated included absence, lateness, derelictions, negative reports, positive reports, assignment to restricted duty, and two supervisory ratings. The subjects were randomly divided into analysis (n=421) and cross-validation samples (n=277). Equations based on the IPI alone accurately identified a minimum of two out of three officers on 75 percent of the performance criteria studied. In contrast, only 38 percent of the MMPI prediction equations reached this rate. Multiple correlations of the IPI with performance ranged from .16 to .34, while MMPI correlations ranged from .05 to .16. Compared to chance classification, the IPI reduced error in "poor" 333 performance predictions in cross-validation samples by 19 percent to 52 percent while the MMPI rates were 8 percent to 38 percent on six of the criteria studied. (Author)