The invention discloses a distributed traffic flow prediction method and system, and the method comprises the steps: 1, obtaining the traffic flow data; 2, processing the traffic flow data by adoptinga pre-established distributed traffic flow prediction model to obtain a prediction result; The generation method of the distributed traffic flow prediction model comprises the following steps: 2.1, calculating a congestion value of an actually measured single road section based on traffic flow data; 2.2, training a single road section traffic flow prediction model based on the actually measured single road section congestion value; And 2.3, adopting a depth random forest, taking the traffic flow data of the selected area and the predicted single-road congestion value output by the single-roadtraffic flow prediction model as training set data, and training the distributed traffic flow prediction model. Calculation is shared by all the nodes for parallel operation, so that the operation amount and the operation time of the server are greatly reduced.