As a result of the limited spatial resolution of positron emission tomographic scanners, the measurements of physiological parameters are compromised by tissue heterogeneity. The effect of tissue heterogeneity on a number of parameters was studied by simulation and an analytical method. Five common tracer models were assessed. The input and tissue response functions were assumed to be free from noise and systematic errors. The kinetic model was assumed to be perfect. Two components with different kinetics were mixed in different proportions and contrast with respect to the model parameters. Different experimental protocols were investigated. Of three methods investigated for the measurement of cerebral blood flow (CBF) (steady state, dynamic, integral), the second one was least sensitive to errors caused by tissue heterogeneity and the main effect was an underestimation of the distribution volume. With the steady state method, errors in oxygen extraction fraction caused by tissue heterogeneity were always found to be less than the corresponding errors in CBF. For myocardial blood flow the steady state method was found to perform better than the bolus method. The net accumulation of substrate (i.e. rCMRglc in the case of glucose analogs) was found to be comparatively insensitive to tissue heterogeneity. Individual rate constants such as k2 and k3 for efflux and metabolism of the substrate in the pool of unmetabolized substrate in the tissue, respectively, were found to be more sensitive. In studies of radioligand binding, using only tracer doses, the effect of tissue heterogeneity on the parameter kon.Bmax could be considerable. In studies of radioligand binding using a protocol with two experiments, one with high and one with low specific activity, Bmax was found to be insensitive while Kd was very sensitive to tissue heterogeneity.