The MMPI-2 profiles of 39 veteran police officers were used to predict their performance (No Apparent Problems, Borderline, or Serious Problems Possible), and these predictions were compared with supervisors' ratings of the officers' actual performance. The MMPI-2 predictions were accurate for 46 percent of the officers, a rate that was significantly better than chance (p = .024). Based on the current data, the best selection outcome would be obtained by accepting officers whose MMPI-2 profiles place them in the "No Apparent Problems" or "Borderline" groups, and rejecting officers whose profiles suggest "Serious Problems Possible." This could be accomplished simply by rejecting any officers who obtained a score above 65T on any of the clinical scales. This selection strategy would have resulted in the acceptance of 22 officers, 19 of whom were highly rated by their supervisors, and the rejection of 17 officers, 11 of whom were rated as borderline or poorer by their supervisors. It would also result in the erroneous rejection of six officers who were highly rated by supervisors. 1 table and 7 references