Transit operations are disturbed frequently by stochastic traffic variations and ridership, which may deteriorate schedule/headway adherence thus lengthen passenger wait times. Providing accurate and accessible information on transit vehicle arrival times is critical to improve transit service quality. In this study, a link-based artificial neural network (ANN) model is developed for predicting bus arrival times in real-time by accumulating travel times on all traversed links between stops. The accuracy of the ANN model is assessed through simulating NJ Transit Route #39 and conducting reliability analysis for predicted bus arrival times. The results show that the model performs well especially with fewer intersections between stops. The study suggests another type of prediction model- stop-based ANN model, which is anticipated to adapt to variation in traffic conditions between stops with more intersections. The study provides an efficient computer program that can be used to integrate and evaluate innovative models (e.g., ANN prediction models) and strategies for promoting service quality.