Five different methods for the estimation of the binding potential, a measure of Bmax/Kd, of [11C]raclopride in human striatum were compared using data from a dose ranging study of the neuroleptic CP-88,059-01. Binding potential was estimated indirectly, from distribution volumes in striatum and cerebellum, using both single- and two-tissue compartment models with a metabolite-corrected plasma curve as input function. The two-tissue compartment model was also used for a direct estimate of the binding potential. In addition, a direct estimate was obtained from the reference tissue compartment model using the cerebellum as indirect input function. Finally, an estimate of binding potential was calculated from the ratio of striatum over cerebellum counts at late times after injection. The estimates of striatum binding potential from all methods, except the direct determination using a two-tissue compartment model with metabolite-corrected plasma input function, correlated with each other. Use of an average metabolite correction resulted in only a small reduction in accuracy in this series of normal subjects. The reference tissue model provided estimates of the binding potential with the same sensitivity for detecting changes as those methods that required a metabolite-corrected plasma input function. This indicates that for routine analysis of clinical [11C]raclopride studies, no arterial cannulation is required. The range of normal values was significantly less variable with the reference tissue method than when simple striatum-to-cerebellum ratios were used.