In mobile and ubiquitous computing the location of devices<br />is often important both for the behavior of the applications<br />and for communication and other middleware functionality.<br />Mobility prediction enables proactively dealing<br />with changes in location dependent functionality. In this<br />project Bayesian networks’ ability to reason on the basis of<br />incomplete or inaccurate information is powering mobility<br />prediction based on a map of the street grid and the current<br />location and direction of the vehicle. We found that it<br />is feasible to divide information of a map into smaller parts<br />and generate a Bayesian network for each of these in order<br />to make mobility prediction based on localized information.<br />This makes the information stored in the Bayesian networks<br />more manageable in size, which is important for resource<br />constrained devices. Common sense knowledge of how vehicle<br />moves is feeded into the networks and enables them<br />to make a good prediction even when no information of the<br />vehicles mobility history is used. Experiments on real world<br />data show that in an area statically divided into hexagonal<br />cells of 200m in diameter, we get 80.54% accuracy when<br />using localized Bayesian networks to predict which cell a<br />vehicle enters next.