The long-term effect of selegiline (L-deprenyl) in the treatment of Parkinson's disease has not been clearly delineated. We report on a group of patients whose treatment was initiated with selegiline (n = 43) and then subsequently included L-dopa-carbidopa (Sinemet) and in whom an extended period of observation was carried out; they are compared to a group of patients whose treatment consisted of L-dopa-carbidopa alone (n = 39). In each, serial observations of the parkinsonian state and the response to treatment on a yearly basis for a period of 5 years were performed. No significant difference in the Hoehn-Yahr stage or in the motor subscores of tremor, rigidity, bradykinesia, and gait-posture was found between the two groups, nor was there a significant difference in the incidence of fluctuating responses or dyskinesias. The group that received combination therapy required less L-dopa than did the group that received L-dopa-carbidopa alone during the first 3 years of treatment and a similar trend was evident in years 4 to 5. We conclude that minimal benefits accrued to the parkinsonian patients from long-term use of selegiline. No clinical evidence to support the claim of "neuroprotective" properties was found. Selegiline's major usefulness is to modify the fluctuating therapeutic response seen with L-dopa-carbidopa.