Utilizing eight apprentices in the mentals trades, this study empirically evaluated a number of techniques designed to produce performance tests that are economically feasible, reliable, content-valid, and less disadvantageous to minority group members than paper-and-pencil measures. The final test was economically feasible, reliable with respect to interscorer agreement and internal consistency, and content valid. In addition, performance test total scores showed substantially smaller majority/minority differences than did paper-and-pencil tests. An adequate test of performance test score stability over time, however, was precluded by practical problems. An additional study was recommended to address the question of temporal stability. Implications for manpower policy, especially in the areas of occupational licensure and apprenticeship training, were discussed. (NTIS/BW)