A prospective trial to demonstrate the efficacy of intrathecal baclofen therapy by implanted pump for adults with spasticity due to spinal cord injury or multiple sclerosis was initiated in our hospital. Of the 140 patients assessed, 7 met the following criteria for inclusion in the study: a modified Ashworth score > 3, a spasm frequency score > 2, and an inadequate response to oral anti-spasticity drugs, (i.e., baclofen, clonidine and cyproheptadine). All patients responded to intrathecal bolus injection of baclofen in the double blind, placebo-controlled screening phase (mean bolus dose = 42.8 micrograms). Programmable Medtronic pumps were implanted in 4 patients while 3 patients received non-programmable Infusaid pumps. Post-implantation, a marked decrease in spasticity occurred with a significant reduction of the Ashworth score (mean = 1.8, p < .005), a reduced spasm score (mean = 0.8, p < .005), and an improved leg swing in the pendulum test. These effects were maintained during a follow-up of 24-41 months (average infusion dose = 218.7 micrograms/day). The gross cost-savings due to reduced hospitalizations related to spasticity was calculated by comparing the cost for the two year period before pump implantation to the same period after treatment for 6 of the 7 patients. The cost of in-hospital implantation as well as the cost of the pumps were deducted from the gross savings. There was a net cost-saving of $153,120. Our findings agree with the reported efficacy and safety of intrathecal baclofen treatment, and illustrate the cost-effectiveness of this treatment.