To clarify the influence of gastric emptying on levodopa-related motor fluctuations in Parkinson's disease, we assessed mobility and plasma levodopa concentrations in 10 patients during five modes of levodopa administration: (1) standard intermittent oral (SIO), (2) intermittent duodenal (ID), (3) continuous duodenal infusion (CDI), (4) continuous gastric infusion (CGI), and (5) controlled-release Sinemet (CR-4). The rank order from greatest to least for both percentage of time "on" and average mobility score was CDI, CGI, ID, CR-4, and SIO. The rank order for variance of means, a measure of fluctuation, from least to greatest for mobility was CDI, CGI, CR-4, ID, SIO, and for plasma levodopa concentrations was CDI, CGI, ID, SIO, and CR-4. The results demonstrate that it is possible to produce very steady plasma concentrations of levodopa with a corresponding reduction in motor fluctuations by continuous intraduodenal administration of the drug. This mode of delivery is an ideal model for the development of optimal continuous-release preparations of levodopa. Other enteral routes have produced a more variable plasma levodopa concentration and clinical response.