Reports on the efficacy of the Minnesota Multiphasic Personality Inventory (MMPI) for selecting patients with valid complaints of pain have been equivocal. The Mensana Clinic Back Pain Test (MPT) was able to predict, with some degree of success, patients who had a definite organic pathologic condition. However, the MMPI measures personality traits, whereas the MPT measures the impact of pain on a patient's life. To determine which of the two tests would be a better predictor of actual physical abnormalities, and hence valid pain complaints, a comparison was undertaken between the two tests. The charts of 83 patients admitted to the Neurosurgery Service of Johns Hopkins Hospital with complaints of back pain were assessed. MMPI test results, as well as test results for the MPT, were compared to the presence or absence of pathologic conditions on electromyography, nerve conduction velocity studies, thermography, myelogram, or computerized axial tomography scan. The MPT had a correlation factor of -.59700, that was significant as P = .000005. Of the 52 patients scoring 17 points or less on the MPT, 85% had objective physical abnormalities, considered moderate or severe by blind review. Of the 31 patients scoring 18 points or greater on the MPT, only 26% had objective physical findings that were considered moderate or severe. Only the F scale (faking badly) of the MMPI correlated with objective physical abnormalities (r = .21340, P less than .033). However, 60% of the patients with T scores of less than 70 on the F scale had objective findings, whereas 75% of patients with T scores greater than 70 had objective physical findings.(ABSTRACT TRUNCATED AT 250 WORDS)