The trapping of decarboxylation products of radiolabelled dopa analogs in living human brain occurs as a function of the activity of dopa decarboxylase. This enzyme is now understood to regulate, with tyrosine hydroxylase, cerebral dopamine synthesis. Influx into brain of dopa decarboxylase substrates such as 6-[18F]fluorodopa and beta-[11C]dopa measured by positron emission tomography can be analyzed by solution of linear differential equations, assuming irreversible trapping of the decarboxylated products in brain. The isolation of specific physiological steps in the pathway for catecholamine synthesis requires compartmental modelling of the observed dynamic time-activity curves in plasma and in brain. The several approaches to the compartmental modelling of the kinetics of labelled substrates of dopa decarboxylase are now systematically and critically reviewed. Labelled catechols are extensively metabolized by hepatic catechol-O-methyltransferase yielding brain-penetrating metabolites. The assumption of a fixed blood-brain permeability ratio for O-methyl-6-[18F]fluorodopa or O-methyl-beta-[11C]dopa to the parent compounds eliminates several parameters from compartmental models. However, catechol-O-methyltransferase activity within brain remains a possible factor in underestimation of cerebral dopa decarboxylase activity. The O-methylation of labelled catechols is blocked with specific enzyme inhibitors, but dopa decarboxylase substrates derived from m-tyrosine may supplant the catechol tracers. The elimination from brain of decarboxylated tracer metabolites can be neglected without great prejudice to the estimation of dopa decarboxylase activity when tracer circulation is less than 60 minutes. However, elimination of dopamine metabolites from brain occurs at a rate close to that observed previously for metabolites of glucose labelled in the 6-position. This phenomenon can cause systematic underestimation of the rate of dopa decarboxylation in brain. The spillover of radioactivity due to the limited spatial resolution of tomographs also results in underestimation of dopa decarboxylase activity, but correction for partial volume effects is now possible. Estimates of dopa decarboxylase activity in human brain are increased several-fold by this correction. Abnormally low influx of dopa decarboxylase tracers in the basal ganglia is characteristic of Parkinson's disease and other movement disorders. Consistent with postmortem results, the impaired retention of labelled dopa is more pronounced in the putamen than in the caudate nucleus of patients with Parkinson's disease; this heterogeneity persists after correction for spillover. Current in vivo assays of dopa decarboxylase activity fail to discriminate clinically distinct stages in the progression of Parkinson's disease and are, by themselves, insufficient for differential diagnosis of Parkinson's disease and other subcortical movement disorders. However, potential new avenues for therapeutics can be tested by quantifying the rate of metabolism of exogenous dopa in living human brain.