Eleven parkinsonian patients participated in a pharmacokinetic/pharmacodynamic study in an attempt to model levodopa (L-DOPA) plasma concentrations to clinical effect. Carbidopa 25 mg/L-DOPA 100 mg (Sinemet 25/100) was given orally, and blood samples were obtained before and serially for 4 hours after the dose. Effect measurements were obtained with each blood sample and included tapping score, timed walking, and global assessment of motor function. Mean L-DOPA plasma concentrations were fitted to a one-compartment pharmacokinetic model. A time-wise plot of modeled plasma L-DOPA concentrations versus mean effect measurements revealed a counter-clockwise hysteresis. Effect compartment concentrations were determined by a least squares approach, which determined elimination rate constants by minimizing hysteresis. Half-times for the equilibration between plasma and the effect compartment were 0.39 h for tapping, 0.36 h for walking, and 0.34 h for the global score. Pharmacodynamic data were fit best with an Emax model with baseline effect for tapping (Emax = 53.2 taps/60 s, EC50 = 0.58 microgram/ml) and global score (Emax set at 5.0 by limits of scale, EC50 = 2.53 micrograms/ml). A linear model best described the relationship between predicted effect site concentration and timed walking. L-DOPA plasma concentrations after oral Sinemet did not correlate well with clinical response because clinical response lags behind plasma concentrations. Half-times for equilibration between plasma and the effect site were similar for all of the effects measured.