In this paper, a trust model is proposed to assess credibility of recommendations in vehicular ad hoc networks (VANETs). In a VANET, nodes share important information with each other. Often these nodes misbehave by sending incorrect information. In order to identify correct information, nodes often use recommendations from their neighbors. However, malicious neighbors may manipulate their recommendations in order to eliminate honest nodes from the network. The trust model provided in this paper will assist nodes to identify such malicious senders and incorrect recommendations. The performance of networks using the proposed trust model is observed to be superior than the existing trust model as suggested by a true positive rate of 0.996 and a false positive rate of 0.001 when malicious senders show selective or probabilistic misbehavior.