We used the ligand 3-N-[2'-18F]fluoroethylspiperone (FESP), which binds to D2-dopamine receptors in the striatum, and positron emission tomography (PET) to quantify striatal D2-dopamine densities (Bmax) and binding kinetics in baboon brain in vivo. Sequential PET scans were obtained for 4 h post injection. Various similar models based on a nonlinear kinetic four-compartment model that takes into account the effect of ligand specific activity were used. We investigated the effect of exact model configuration on the reliability of Bmax and other kinetic transfer coefficients. We found that with the ligand FESP and dynamic PET studies, the estimated values of Bmax and other model parameters are sensitive to the choice of model configuration, ligand specific activity, and data analysis technique. The limitations of the reliability of parameter estimates in a complex kinetic model for receptor ligands were studied in simulation calculations. Results showed that the accuracy of estimated values of Bmax is affected by both the ligand binding properties and the injected dose of ligand. The estimated average value of kinetic model parameters was as follows: ligand-receptor dissociation constant k4 = 0.0080 min-1; the product of ligand-receptor association constant and fraction of ligand available to bind to specific receptors f2ka = 0.0052 (min nM)-1; and D2-dopamine receptor density Bmax = 37.5 pmol g-1.