The objective of this study was to examine the effect of long-term treatment with selegiline on the progression of Parkinson's disease (PD). One hundred and sixty-three patients with early PD were treated with levodopa and benserazide, combined with selegiline or placebo in a five-year randomized, placebo-controlled, double-blind, parallel group study followed by a one-month wash-out of selegiline or placebo. The main outcome measures were assessments of the severity of parkinsonism, levodopa requirements and the development of end-of-dose motor fluctuations over time and after wash-out at the end of the study period. Results indicated that patients treated with the combination of selegiline and levodopa developed markedly less severe parkinsonism and required lower doses of levodopa during the five-year study period than patients treated with levodopa and placebo. There was no trend towards worsening during wash-out among patients previously treated with selegiline. The results cannot easily be explained by a symptomatic effect of selegiline.