Spinal cord stimulation, in use for more than 20 years, has evolved into an easily implemented technique, with percutaneous methods for electrode placement. We have reviewed our experience with this technique in treating "failed back surgery syndrome," and have assessed patient and treatment characteristics as predictors of long-term outcome. A series of 50 patients with failed back surgery syndrome (averaging 3.1 previous operations), who underwent spinal cord stimulator implantation, was interviewed by impartial third parties, at mean follow-up intervals of 2.2 years and 5.0 years. Successful outcome (at least 50% sustained relief of pain and patient satisfaction with the result) was recorded in 53% of patients at 2.2 years and in 47% of patients at 5.0 years postoperatively. Ten of 40 patients who were disabled preoperatively returned to work. Improvements in activities of daily living were recorded in most patients for most activities; loss of function was rare. Most patients reduced or eliminated analgesic intake. Statistical analysis (including univariate and multivariate logistic regression) of patient characteristics as prognostic factors showed significant advantages for female patients and for those with programmable multi-contact implanted devices. These results, in patients with postsurgical lumbar arachnoid and epidural fibrosis and without surgically remediable lesions, compare favorably with the results in two separate series of patients with failed back surgery syndrome, in whom 1) surgical lesions were diagnosed and repeated operation performed; and 2) monoradicular pain syndromes were diagnosed and dorsal root ganglionectomies performed at our institution. This suggests the need for further assessment of selection criteria, critical analysis of treatment outcome, and prospective study of spinal cord stimulation and alternative approaches to failed back surgery syndrome.