Although the criterion problem has been acknowledged as critical in personnel research, few attempts have been made to systematically examine the nature and covariates of criterion measures of performance. The present research used meta-analytic techniques to examine the race effect size for objective measures of performance and to compare the relationship between effect sizes for objective indices and subjective ratings. Fifty-three studies were located that included at least one objective index of actual performance, absenteeism or cognitive test performance and one subjective measure of performance for the same group of black and white employees. The corrected average effect sizes across the 53 studies were relatively low but quite similar for the objective and subjective criteria. Moderating effects for the objective criteria were found as race effects were much higher for cognitive than for performance criteria. Subjective ratings had a lower effect size than objective cognitive test scores but were higher than comparable objective performance indices. The implications of the results for personnel research practices were discussed and the need for a better understanding of the constructs underlying criterion measures were emphasized. (Author)