The invention discloses an intersection signal parameter optimization method based on Monte Carlo tree search and can optimize signal timing of an intersection of the urban road network in real time.The method specifically comprises steps that firstly, an candidate phase set and a performance function for each candidate phase are defined, and secondly, signal parameters within the optimized timewindow are continuously searched. Each search comprises (1), phase sampling; (2), traffic flow simulation; and (3), performance function update. Through several searches, based on the updated performance functions, the best performance of each candidate phase and the corresponding phase duration are obtained, and lastly, the best with the best phase performance and the corresponding phase durationare selected as the next phase scheme. The method is advantaged in that requirements for traffic flow models are not high, universality is strong, the phase schemes can be flexibly selected, and pre-setting of the phase and the phase sequence is not required.