In order to address the problem of directly discharging of untreated sewage from storm pipes causing a range of water pollution problems and the uncertainty problem of source inference due to the complexity of drainage system conditions, the storm water management model (SWMM) was employed to simulate the variation of water quality within the drainage system. A SWMM-Bayesian traceability model based on the Bayesian-MCMC (Monte Carlo Markov Chain) method was constructed using the MatSWMM toolbox to predict the probability distributions of the location of each potential source, the extent of discharge, and the time of discharge. Example validation results show that in the case of reasoning about a single unknown parameter of a single pollution source, the SWMM-Bayesian traceability model can accurately determine the true value of the unknown parameter. When all three parameters of a single pollution source are unknown, due to the existence of a variety of combinatorial properties, the traceability accuracy will be reduced significantly, and can only be judged to be in the approximate range, but it can be increased by adding the appropriate water quality monitoring points, to improve the accuracy and efficiency of SWMM-Bayesian traceability model. For the case that the maximum number of pollution sources is 2, the SWMM-Bayesian traceability model is easy to be trapped in the local optimal solution, and the improvement of the traceability model by using the likelihood function to optimize the initial value of the iterative process can effectively solve the problem of local optimal solution.