Concerns about exposure assessment quality have impeded research to identify risk factors for ergonomic disorders. We compared self-reported and expert-observed estimates of work-related physical factors for participants in a study of carpal tunnel syndrome (CTS). We analyzed data from 61 subjects, including 28 CTS cases and 33 controls randomly samples from a case-control study with 417 participants. For 11 posture and manual materials handling factors, the median difference in mean exposure between self-reported and expert-observed exposure at work was less than 1/2 hour a day. Measurements by the two methods in this study agreed more often than expected by chance (median kappa 0.31 in cases and 0.28 in controls). Kappa differed significantly by case-control status for two factors: bending at the waist (kappa 0.79 in cases versus 0.28 in controls, P = 0.01) and twisting of the forearm (kappa 0.45 in cases versus -0.02 in controls, P = 0.02). Although imperfect, exposure information collected from workers' self-reports is useful for many ergonomic epidemiology studies.