Spinal cord stimulation has evolved over the past 20 years into an easily implemented technique, with low morbidity, for the treatment of intractable, chronic pain in properly selected patients. We report our experience with a series of 62 patients implanted between 1983 and 1987, with percutaneous and laminectomy electrodes, and with single- and "multi-channel" (programmable, multi-contact) devices. Fifty had chronic, intractable low back and leg pain ("failed back surgery syndrome," lumbar arachnoid fibrosis), five had spinal cord injuries, and seven "peripheral" pathology or stump pain. Statistical analysis of these and other patient characteristics and technical factors was undertaken to identify predictors of outcome. All patients were interviewed by a disinterested third party at a mean of 2.14 years following implantation. A majority of patients reported at least 50% sustained relief of pain and indicated that they would go through the procedure again for the same result. There was corresponding improvement in ability to perform various everyday activities, and decrease in use of analgesics. Ten of 40 failed back patients who were disabled before the procedure returned to work postoperatively. Superposition of stimulation paresthesias upon a patient's topography of pain was found to be a statistically significant predictor of successful relief of pain, by linear regression methods. Univariate and multivariate analysis of patient characteristics and technical factors as predictors of outcome demonstrated significant advantages for female patients, and for patients implanted with "multi-channel" devices. With these devices, electrode geometries with central cathode(s) flanked by rostral and caudal anode(s) were favored disproportionately. Technical improvements in implanted spinal cord stimulation devices, in particular the development of multi-contact percutaneous electrode arrays and supporting programmable electronics, have significantly improved clinical results.