The invention provides a method and device for parameter automatic calibration of a rail transit passenger flow distributing model, and belongs to the technical field of intelligent transportation. According to the method, the genetic algorithm is adopted for executing the passenger flow distributing model, the nonparametric statics inspection technology is used for inspecting selected model parameter values, and model parameter values which pass through inspection are used as calibration results. The device comprises a flow allocation model calibration module, a road network data managing module, an automatic ticket selling and checking data storing module, an operation plan managing module and a network passenger flow distributing module, wherein the flow allocation model calibration module is connected with the road network data managing module, the automatic ticket selling and checking data storing module, the operation plan managing module and the network passenger flow distributing module. The method and device can improve accuracy of the rail transit passenger flow distributing model; model parameter calibration can be carried out based on urban rail transit system AFC data, and therefore waste of labor, material and financial resources due to the fact that a manual passenger flow investigation method is frequently used is avoided; due to the fact that the device can be integrated in an urban rail transit system as an automatic calibration module, the device can promote improvement of urban rail transit operation managing level.