Levodopa has several advantages as a pharmacological challenge agent for human neuroscience research. Exogenous levodopa changes striatal neuronal activity and increases extracellular dopamine concentrations, and with adequate inhibition of peripheral metabolism levodopa does not change mean cerebral blood flow. For neuroimaging studies of Parkinson disease (PD) and Tourette syndrome, we sought to rapidly produce a biologically relevant steady-state levodopa concentration and then maintain that concentration for at least an hour. We also wished to minimize side effects, even in individuals without prior levodopa treatment. We designed a two-stage intravenous infusion protocol based on published levodopa pharmacokinetic data. We report results of 125 infusions in 106 subjects, including healthy volunteers, PD patients, and people with chronic tics. At higher doses (target steady-state levodopa concentrations of 2,169 and 1,200 ng/ml), treatment-naive volunteers had unacceptably frequent side effects. The final infusion protocol, with a target steady-state concentration of 600 ng/ml, was well-tolerated (mild nausea in 11% of subjects was the only side effect occurring significantly more than in single-blind saline infusions), produced the desired plasma levodopa concentration (612+/-187 ng/ml, mean+/-S.D.), and produced statistically significant antiparkinsonian benefit (16% mean reduction in a standard rating of parkinsonian motor signs, P<0.0005).