Vehicular Ad hoc Networks (VANETs) are a special case of ad hoc networks with highly mobile nodes (on-board vehicles). One of the goals of VANETs is to use trusted relationships among vehicles to achieve cooperative driving. In this paper, we present a trust management system that uses similarity to establish and maintain trust relationships among the vehicles in the network. Additionally, we present a scheme that uses the similarity-based trust relationships to detect false safety event messages from abnormal vehicles in VANETs. Our scheme responds to safety events claims made by a vehicle by predicting the behavior of the reporting vehicle. We perform simulations to study the effectiveness of the proposed scheme in uncovering false safety event messages sent by abnormal vehicles.