As the number of various positioning sensors and location-based devices increase, a huge amount of spatial and temporal information data is collected and accumulated. These data are expressed as trajectory data by connecting the data points in chronological sequence, and thses data contain movement information of any moving object. Particularly, in this study, urban vehicle trajectory prediction is studied using trajectory data of vehicles in urban traffic network. In the previous work, Recurrent Neural Network model for urban vehicle trajectory prediction is proposed. For the further improvement of the model, in this study, we propose Attention-based Recurrent Neural Network model for urban vehicle trajectory prediction. In this proposed model, we use attention mechanism to incorporate network traffic state data into urban vehicle trajectory prediction. The model is evaluated by using the Bluetooth data collected in Brisbane, Australia, which contains the movement information of private vehicles. The performance of the model is evaluated with 5 metrics, which are BLEU-1, BLEU-2, BLEU-3, BLEU-4, and METEOR. The result shows that ARNN model have better performance compared to RNN model.