Over the past two decades, spinal cord stimulation devices and techniques have evolved from single-channel systems, with electrodes requiring laminectomy, into programmable "multichannel" systems with electrodes that may be placed percutaneously. We have reviewed our experience in 320 consecutive patients treated with these devices at our institution between 1972 and 1990. Technical details of treatment as well as patient characteristics have been assessed as predictors of clinical outcome and of hardware reliability by univariate and multivariate statistical methods. Current follow-up has been obtained at intervals from 2 to 20 years (mean, 7.1 yr) postoperatively on 205 patients. All clinical outcome measures have been based on disinterested third-party interview data--standard analog pain ratings, employment status, activities of daily living, and use of analgesics. At 7-year mean follow-up, 52% of the 171 patients who received permanent implants reported at least 50% continued pain relief. A majority had maintained improvements in activities of daily living and analgesic use. Analysis of hardware reliability for 298 permanent implants revealed significantly fewer clinical failures (P < 0.001) and technical failures (in particular, electrode migration and malposition, P = 0.025) as single-channel implants have evolved into programmable, multichannel devices. Our analysis of technical and clinical prognostic factors may be useful to the clinician in selecting patients for this procedure.