In 1993, the last AAN Practice Parameter on medical treatment of Parkinson's disease (PD) concluded that levodopa was the most effective drug for management of this disorder. Since then, a number of new compounds including non-ergot dopamine agonists (DA) and sustained-release levodopa have been released and studied. Thus, the issue of treatment in de novo PD patients warrants reexamination. Specific questions include: 1) does selegiline offer neuroprotection; 2) what is the best agent with which to initiate symptomatic treatment in de novo PD; and 3) is there a benefit of sustained release levodopa over immediate-release levodopa? Using evidence-based principles, a literature review using MEDLINE, EMBASE, and the Cochrane Library was performed to identify all human trials in de novo PD between 1966 and 1999. Only articles that fulfilled class I or class II evidence were included. Based on this review, the authors conclude: 1) Selegiline has very mild symptomatic benefit (level A, class II evidence) with no evidence for neuroprotective benefit (level U, class II evidence). 2) For PD patients requiring initiation of symptomatic therapy, either levodopa or a DA can be used (level A, class I and class II evidence). Levodopa provides superior motor benefit but is associated with a higher risk of dyskinesia. 3) No evidence was found that initiating treatment with sustained-release levodopa provides an advantage over immediate-release levodopa (level B, class II evidence).