The association between manual materials handling and occupational low back injuries has generated a wide variety of devices and protocols for testing lifting capacity. The validity of these tests depends critically on the subject's effort during the evaluation. This study compares the accuracy of several physiologic effort indices in identifying maximal and submaximal lifting efforts. Twenty-one men and twenty-three women free from back pain underwent isometric, isokinetic, and isoinertial tests of lifting capacity applying 50 and 100% efforts in random order. Effort indices included isokinetic force/distance curve variation, isoinertial peak force: weight ratios and peak force--weight differences, and heart rates and peak force variances for all three modes. Differences between 50 and 100% effort means were significant at P less than 0.01 for isokinetic force/distance curve variation, isoinertial peak force: weight ratios, and peak force--weight differences, and for heart rates in all three modes using analysis of variance. Differences were not significant (P greater than 0.05) for peak force variances in any mode using Wilcoxon Signed Rank Tests. According to a discriminant analysis model with optimal cutoff values, the only effort indices with better than 60% accuracy in identifying maximal and submaximal efforts were isokinetic force/distance curve variation (74%) and isoinertial peak force: weight ratios (69%), peak force--weight differences (62%) and heart rates (65%). For each index, subjects were divided into three groups according to whether their lifting outputs during 50% effort bouts were greater than, equal to, or less than one half their outputs in the 100% effort bouts.(ABSTRACT TRUNCATED AT 250 WORDS)