With the advancement of mobile computing technology and the widespread use of GPS-enabled mobile devices, the location-based services have received more and more attentions, and the path prediction of moving objects is one of the most important issues. The existing prediction methods of moving objects focus mainly on the precise historic trajectory in Euclidean space. However, in the real world, there are a lot of applications which require predicting network-constrained trajectory based on the uncertain historic trajectory. As yet, there has been no research on uncertain path prediction of moving objects on road networks. In order to solve this problem, a method of generating the uncertain trajectory is proposed firstly, the definition of path probability and an uncertain path prefix tree are used to generate the uncertain trajectory, and a corresponding data format of the uncertain trajectory is given. Then an uncertain trajectory pattern mining algorithm is proposed, and a data structure named id-list is used in the algorithm. Finally trajectory patterns which are mined from the uncertain trajectory pattern mining algorithm are indexed by a novel access method for efficient query processing. The experiment shows good performance of the system, and the results demonstrate that the proposed techniques are accurate, efficient and of low storage capacity.