Biochemical approaches were used in freely moving rats to determine, under steady-state conditions, the brain/arterial plasma partition coefficients of L-tryptophan and alpha-[3H]methyl-L-tryptophan, from which the lumped constant for the alpha-methyl-L-tryptophan method of estimating the rate of brain serotonin synthesis is calculated. The lumped constants were significantly different in the various structures examined: 0.149 +/- 0.003 in the raphe dorsalis, 0.103 +/- 0.002 in the raphe centralis, 0.087 +/- 0.003 in the reticular formation, and 0.62 +/- 0.08 in the pineal gland. From these data we proposed a two-compartment model to calculate the rate of serotonin synthesis by quantitative autoradiography using a three-time point experiment. Rates of synthesis for the raphe dorsalis and the reticular formation (620 +/- 57 and 80 +/- 35 pmol/g of tissue/min, respectively) were similar to those measured simultaneously by biochemical means, but rates were 50% higher for the raphe centralis (568 +/- 90 vs. 381 +/- 31 pmol/g of tissue/min). The lack of dynamic equilibrium of the tracer between plasma and tissue pools may explain the discrepancy between the two methods. Our findings did not confirm previous data, indicating that the application of the autoradiographic method to measure the rate of brain serotonin synthesis using alpha-methyl-L-tryptophan as tracer has limitations.