To assess the stability of different measures of receptor occupancy from [11C]raclopride (a D2 antagonist) studies with positron emission tomography, we analyze data from five test/retest studies in normal volunteers in terms of individual model parameters from a three-compartment model, the distribution volume (DV) and the ratio of DVs from a receptor-containing region of interest to a non-receptor-containing region. Large variations were found in the individual model parameters, limiting their usefulness as an indicator of change in receptor systems. The DV ratio showed the smallest variation. Individual differences were reflected in the greater intersubject variation in DV than intrasubject variation. The potential effects of blood flow on these measurements were addressed both experimentally and by simulation studies using three models that explicitly incorporate blood flow into a compartmental model that also includes receptor-ligand binding. None of the models showed any variation in the DV with changes in blood flow as long as flow was held constant during the simulation. Experimentally, blood flow was significantly reduced by hyperventilation in a human subject. The DV was found to be reduced relative to baseline in the hyperventilation study, but the DV ratio remained unchanged. The effect of elevated and reduced flow was also tested in two baboon experiments in which PCO2 was varied. Some variability in the DV ratio was observed but was not correlated with changes in blood flow. This raises the possibility that other factors indirectly related to changes in blood flow (or PCO2) may cause changes in DV, and these effects need to be considered when evaluating experimental results.