The biological accuracy of a nonlinear compartmental model describing the in vivo kinetics of L-3,4-dihydroxy-6-[18F]fluorophenylalanine ([18F]FDOPA) metabolism was investigated. Tissue activities for [18F]FDOPA and its labeled metabolites 3-O-methyl-[18F]FDOPA ([18F]OMFD), 6-[18F]fluorodopamine ([18F]FDA), L-3,4-dihydroxy-6-[18F]fluorophenylacetic acid ([18F]FDOPAC), and 6-[18F]fluorohomovanillic acid ([18F]FHVA) were calculated using a plasma [18F]FDOPA input function, and kinetic constants estimated previously by chromatographic fractionation of 18F-labeled compounds in plasma and brain extracts from rat. Present data accurately reflected the measured radiochemical composition in rat brain for tracer circulation times past 10 min. We formulated the hypothesis that the discrepancy between calculated and measured fractions of [18F]FDOPA and the deaminated metabolite [18F]FDOPAC at times earlier than 10 min reflected storage of [18F]FDA in vesicles without monoamine oxidase. This hypothesis explained the initially rapid appearance of [18F]FDOPAC in striatum by delayed transfer of [18F]FDA from cytosol into vesicles. We conclude that the simpler model of [18F]FDOPA compartmentation is accurate when the cytosolic and vesicular fractions of [18F]FDA are at steady-state; the approach to equilibrium has a time constant of 15-30 min. The present model is valid for positron emission tomography studies of [18F]FDOPA metabolism in living brain.