The effect of pergolide 25 mug twice daily on levodopa initiation was assessed in a randomized, placebo-controlled, parallel group, double-blind multicenter trial in 106 untreated early Parkinson's disease patients. The primary endpoint of mean time until levodopa was 520 days (95% confidence interval [CI], 422-618 days) for pergolide versus 434 days (95% CI, 358-609 days) for placebo. However, this increase of 86 days for pergolide was not statistically significant. The wash-in effect of pergolide was significant at 6 weeks (change in mean Unified Parkinson's Disease Rating Scale [UPDRS] 2 and 3 was -0.1 [95% CI, -1.4 to 1.3] for pergolide vs. 2.2 [95% CI, 1.1-3.3] for placebo). At termination, the change from baseline in mean UPDRS 2 and 3 score was 11.4 (95% CI, 8.8-14) for pergolide and 14.6 (95% CI, 12-17.2) for placebo (P=0.08). There was no significant change in UPDRS 2 and 3 for the 83 patients achieving the planned 4-week washout at termination (pergolide 1.2 [95% CI, -0.8 to 3.2] vs. placebo 0.0 [95% CI, -1.6 to 1.6]. Adverse events were infrequent and occurred equally for pergolide and placebo. The study shows no evidence of a neuroprotective effect but indicates a mild symptomatic benefit from pergolide at a dose normally considered subtherapeutic.