A total of 93 patients with intractable spasticity due to either spinal cord injury (59 cases), multiple sclerosis (31 cases), or other spinal pathology (three cases) were entered into a randomized double-blind placebo-controlled screening protocol of intrathecal baclofen test injections. Of the 88 patients who responded to an intrathecal bolus of 50, 75, or 100 micrograms of baclofen, 75 underwent implantation of a programmable pump system for chronic therapy. Patients were followed for 5 to 41 months after surgery (mean 19 months). No deaths or new permanent neurological deficits occurred as a result of surgery or chronic intrathecal baclofen administration. Rigidity was reduced from a mean preoperative Ashworth scale score of 3.9 to a mean postoperative score of 1.7. Muscle spasms were reduced from a mean preoperative score of 3.1 (on a four-point scale) to a mean postoperative score of 1.0. Although the dose of intrathecal baclofen required to control spasticity increased with time, drug tolerance was not a limiting factor in this study. Only one patient withdrew from the study because of a late surgical complication (pump pocket infection). Another patient received an intrathecal baclofen overdose because of a human error in programming the pump. The results of this study indicate that intrathecal baclofen infusion can be safe and effective for the long-term treatment of intractable spasticity in patients with spinal cord injury or multiple sclerosis.