Intelligent routing systems, for example, regulate the traffic flow based on information about the traffic state in a defined neighbourhood. Traffic state identification can be realised by identifying and tracking each vehicle in this neighbourhood. Another example could be passenger information systems, that are able to e.g. estimate arrival times on bus stops based on a measurement of the vehicle's position. There are different sources from which the required positioning data are available, e.g. as a result of traffic-monitoring systems or GPS data of vehicle fleets. These sources usually provide sets of noisy or incomplete measurements. In order to calculate potentially optimal state estimates from distorted measurements one has to make use of suitable stochastic filtering methods. Applying a Kalman filter (KF) to the tracking problem yields optimal state estimates of a tracked vehicle in case certain assumptions concerning the linearity of the underlying kinematic model and the noise processes (white gaussian) hold. Particle filters (PFs), on the other hand, are also suitable for (highly) nonlinear systems and white noise but yield only sub optimal state estimates. In this paper we address the problem of applying PFs to the problem of object tracking, taking advantage of the specific properties of this approach. This paper examines differences between the two approaches with respect to practical application for tracking vehicles. For instance particle filters are an ideal counterpart of top-down approaches for finding measurements. The authors also discuss other properties of PFs such as the possibility of tracking multiple hypotheses and processing of ambiguous measurements and evaluate their usefulness for vehicle tracking. The paper is structured as follows. Chapter 2 contains a short introduction to the problem of tracking vehicles in video sequences and the fundamentals of Bayesian state estimation as a common foundation of both, the KF and the PF. In Chapter 3 the concept of Particle filters is being introduced and in chapter 4 the authors discuss aspects of the practical application of a PF to the problem of vehicle tracking. Conclusions are given in chapter 5.