Nowadays, vehicle industrialization has realized several connectivity protocols to enable in-vehicle network communication. These protocols have been collectively standardized in a de facto standard for the in-vehicle network viz controller area network (CAN). Merely, CAN protocol shortages several security features that make vehicular communications susceptible to diverse message injection attacks that may mislead original electronic control units (ECUs) or cause failures. Therefore, defending the in-vehicle network from cyber-attacks is an essential concern. This paper proposes a fast anomalous traffic detection system for secure vehicular communications. The proposed system differentiates the performance of four different machine-learning approaches: Adaboost trees (ABT), Coarse decision trees (CDT), naïve Bayes classifier (NBC), and support vector machine (SVM). The models were evaluated on a recent dataset from a real-time vehicular communications environment, the car-hacking-2018 dataset. Specifically, the system considers five balanced classes, including one normal traffic class and four classes for message injection attacks over the in-vehicle controller area network: fuzzy attack, DoS attack, RPM attack (spoofing), and gear attack (spoofing). Our best performance outcomes belong to the ABT model, which notched 99.8% classification accuracy and 6.67 µseconds of classification overhead. Such results have outweighed existing in-vehicle intrusion detection systems employing the same/similar dataset.