Two hundred patients at a median age of 63 years, receiving conventional levodopa therapy for 8 years, who had had Parkinson's disease for 10 years, tried a regimen of selegiline (L-deprenyl), a type B monoamine oxidase inhibitor, at a daily dose of 10 mg, for varying periods from less than 6 months to more than 24 months (28% over 24 months). Selegiline does improve parkinsonism during the initial 6 months to 12 to 24 months of combined therapy in one third to almost half of patients with an end-of-dose type of response to long-term levodopa therapy. However, even this particular class of patients is unable to maintain such an improvement by 36 months, much less by 48 months, from the start of the selegiline trial. About one quarter of poor responders to levodopa and those with random deterioration show improvement in their parkinsonian status in the first 6 months of the selegiline trial, but their conditions quickly deteriorate by 1 year. The predominant pattern of response to previous levodopa therapy and the severity of the total disability score at the initiation of the selegiline trial were the two variables that were predictive of risk of failure with the drug. No evidence suggested that selegiline decreases the excess mortality rate of Parkinson's disease above that achieved with the use of levodopa alone. Selegiline as an adjunctive agent to conventional levodopa therapy was not unduly impressive with regard to preventing progression of Parkinson's disease.