Fifty-four patients treated in a three-week in-patient rehabilitation program were randomly assigned to and accepted treatment with electroacupuncture (n = 17), TENS (low-intensity transcutaneous nerve stimulation, n = 18) and TENS-dead battery (placebo, n = 18). Outcome measures included estimates of pain (on a visual analogue scale) and disability by both physician and patient as well as physical measures of spine function. Two groups were constructed based on the absence of nonorganic physical findings (Valid group, n = 30) and the presence of two or more nonorganic physical findings out of a possible four (Invalid group, n = 10). Multivariate and univariate analyses of covariance were utilized to determine effects of treatment (acupuncture, TENS, placebo) and the effects of over-reporting (presence of excessive nonorganic physical findings). Statistically significant findings demonstrated that the acupuncture group enjoyed more relief of peak pain and more relief of pain on an average day at the three-month return assessment. Additionally, the acupuncture group demonstrated greater improvement in extension trunk strength at the discharge assessment. The Invalid group were found to have a contaminating effect on the acupuncture results. Analysis also demonstrated associations between nonorganic physical findings and both personality traits (“Conversion V” profile on MMPI) and retention of an attorney. Researchers conducting clinical trials in chronic low-back pain patients should control for contamination by the presence of overreporters.