We studied the effect of the intrathecal infusion of baclofen, an agonist of gamma-aminobutyric acid, on abnormal muscle tone and spasms associated with spinal spasticity, in a randomized double-blind crossover study. Twenty patients with spinal spasticity caused by multiple sclerosis or spinal-cord injury who had had no response to treatment with oral baclofen received an intrathecal infusion of baclofen or saline for three days. The infusions were administered by means of a programmable pump implanted in the lumbar subarachnoid space. Muscle tone decreased in all 20 patients (mean [+/- SD] Ashworth score for rigidity, from 4.0 +/- 1.0 to 1.2 +/- 0.4; P less than 0.0001), and spasms were decreased in 18 of the 19 patients who had spasms (mean [+/- SD] score for spasm frequency, from 3.3 +/- 1.2 to 0.4 +/- 0.8; P less than 0.0005). Tests for motor function, neurologic examination, and assessments by the patients correctly indicated when baclofen was being infused in all cases. All patients were then entered in an open long-term trial of continuous infusion of intrathecal baclofen. During a mean follow-up period of 19.2 months (range, 10 to 33), muscle tone has been maintained within the normal range (mean Ashworth score, 1.0 +/- 0.1) and spasms have been reduced to a level that does not interfere with activities of daily living (mean spasm score, 0.3 +/- 0.6). No drowsiness or confusion occurred, one pump failed, and two catheters became dislodged and had to be replaced. No infections were observed. Our observations suggest that intrathecal baclofen is an effective long-term treatment for spinal spasticity that has not responded to oral baclofen.