We develop an algorithm for predicting the arrival time of a transit vehicle at a signalised intersection, with specific application for signal priority control. In order to accurately predict the arrival time, real-time GPS bus location information is made available. We propose an algorithm that uses both realtime and historical vehicle location information for predicting the arrival time at the next traffic light. The predictor is consisted of two models: (i) a historical model and (ii) an adaptive model that adaptively adjusts its filter gain based on the real-time data. The estimates generated by these two models are molded in a weighted average. The prediction algorithm adaptively adjusts its weight distribution according to some error variances obtained from the two models.