A prospective, blinded cohort study was performed to investigate the relationship between biomechanical variables measured during lumbar dynamometry and several psychological tests and measures of nonorganic pain behavior. Eighty-five men, aged 18-60 years, who had had low-back pain for longer than 5 weeks participated in the study. Nonorganic pain behavior was measured with the Waddell score, and lumbar function was measured with the Isostation B-200 Lumbar Dynamometer. Two brief psychological tests, the Coopersmith Self-Esteem Inventory and an analog self-rating of wellness, were also administered. Relationships between biomechanical variables and psychological tests were calculated with thet-test, the Pearsonrcorrelation, analysis of variance, and multiple step-wise logistic regression. Patients who exhibited excessive illness behavior (Waddell Scores 3-5) performed significantly worse (P< .01) on almost all biomechanical variables. The Coopersmith Self-Esteem Score and “feelings” score had a slightly weaker but still significant correlation with motor performance. The results suggest that poor performance on biomechanical testing in this population may be a form of abnormal illness behavior and thus may not accurately reflect organic alterations of neuromusculoskeletal function.