In order to fully and accurately analyze the spatio-temporal distribution of link travel time,this paper incorporates the time sequence and the spatial relationship between two links into two adjacent links for travel time reliability prediction.From the temporal dimension,the widely-used Kalman filtering is adopted to predict the travel time; while from the spatial dimension,the correlation model describing the travel time between downstream and upstream links is built based on the discrete Markov chain model.Then,a spatio-temporal Bayesian model ( STBM) is constructed to fuse the temporal and spatial travel time distributions for predicting the travel time reliability of road links.Case study results show that,as compared with the prior distributions,the proposed model decreases the reliability prediction error by 45. 7% and 29. 2% respectively for the two measured adjacent links,which verifies the effectiveness of the ST-BM model.