We studied survival in all Olmsted County Parkinson's disease (PD) patients seen at the Mayo Clinic from 1964 to 1978, attempting to answer two questions: (1) What effect does levodopa have on survival in PD? and (2) Does the timing of levodopa administration influence survival? We chose this period because it allowed us to study patient records with a spectrum of disease durations before levodopa treatment; in many patients, the treatment delay was exclusively due to levodopa being unavailable prior to 1969. Mortality of the entire PD cohort (N = 179; 61% levodopa-treated) was greater than that of the general population (matched chronologically, geographically, and by age and gender). Lower age at onset of motor symptoms, lower Hoehn and Yahr stage at first neurologic visit for parkinsonism, and treatment with levodopa were all independent predictors of improved survival. Using a time-dependent Cox regression model, we assessed the impact of the timing of levodopa administration during the course of illness on mortality, while statistically controlling for other factors (ie, patient selection for levodopa treatment, and independent predictors of survival). Risk of death following initiation of levodopa was significantly reduced (p < 0.001), regardless of pre-levodopa duration of illness. This reduction gradually diminished over a period of 4 years on levodopa, but continued to be significantly reduced. After 4 years, increasing survival benefit again progressively accrued over time to at least 17 years of levodopa treatment (p < 0.001). At no point in time was levodopa treatment associated with increased mortality, arguing against substantial levodopa toxicity. However, despite levodopa-improved survival, mortality continues to be increased in PD relative to the general population.