Acupuncture treatment of chronic low back pain was studied in a placebo-controlled double-blind crossover trial completed by 77 patients. The patients had significantly increased depression, neuroticism, and hypochondriasis scores. Initial pain levels correlated with state-anxiety, depression, pain duration, and abnormal illness behavior measures, as well as with the intake of psychotropic but not analgesic medication. Overall reduction in pain score was 26 percent for acupuncture and 22 percent for placebo treatment; the difference was not significant (p greater than 0.6). Analgesic drug intake was reduced to a similar extent in both groups. During the first phase of treatment, patients receiving acupuncture had a greater but not significantly different reduction in pain rating scores compared with those receiving placebo (t = 0.52; p greater than 0.6). This group showed significantly lower pain scores (p less than 0.05) in the second phase of the trial while receiving placebo treatment. Overall reduction in individual patient's pain score was best predicted by initial pain severity (r = 0.43; p less than 0.001) and psychotropic drug intake (r = 0.37; p less than 0.001). None of the variables tested predicted which patients would specifically respond to acupuncture or placebo.