This paper deals with a novel sensor fusion approach to detect and track cars and pedestrians to facilitate a collision mitigation application for vehicles. Robust collision mitigation requires a perception performance of an unprecedented degree of reliability, since an erroneous application of emergency braking caused by false alarms would greatly impede road safety improvement not lastly due to the major setback such an incident would represent for driver acceptance. However, current off-the-shelf single sensor approaches can hardly fulfil the challenging demands. Accordingly, we develop a multi-sensor recognition system. It is composed of a far infrared imaging device, a laser scanner and several radar sensors, which operate integrated into a BMW sedan.