The invention discloses an automatic parameter search method for an McMaster event detection algorithm. The method comprises the steps that S1, on the premise of acquiring a large amount of traffic data, statistical analysis is performed on flow and occupancy respectively to determine a traffic flow distribution model and an occupancy distribution model; S2, a maximum likelihood accounting method is adopted to estimate parameters of the flow distribution model and the occupancy distribution model; S3, quantiles are calculated to determine an approximate interval of flow-occupancy turning points; and S4, the optimal turning point within the turning interval is searched for. Through the method capable of being used for automatically acquiring McMaster parameters, manual dependency during parameter selection is lowered, and the method plays an important role in improving the accuracy and transportability under different road segments of the McMaster algorithm.