Maps of perfusion and Arterial Transit Time (ATT) can be measured quantitatively using non-invasive Pulsed Arterial Spin Labeling (PASL) techniques. This can be achieved by fitting a kinetic model to magnetization difference data images acquired at multiple inversion times (TI). Here, spatial information is incorporated into a Bayesian estimation method, based on a maximum a posteriori (MAP) criterion, which also incorporates a priori knowledge regarding the model physiological parameters. Two types of spatial priors were tested. Monte Carlo simulations showed reduced parameter estimation errors when including spatial information using a Total Variation prior. Furthermore, the feasibility of the method proposed here was demonstrated through the application to empirical data.