The invention provides a regional long-term inflowing multi-target joint probability prediction method. The method comprises: establishing a layered Bayesian probabilistic model with consideration of independent characteristics and population characteristics of a plurality of predicted targets; on the basis of sample data of the predicted targets, carrying out an analysis to obtain predicted target prior distribution, carrying out sampling on independent characteristic parameters and population characteristic parameters of the predicted targets by using a Markov chain Monte Carlo method, optimizing model parameters of the layered Bayesian probabilistic model, and calculating a predicted target plan set meeting conditions. The regional long-term multi-target joint probability prediction method has the following advantages: the layered Bayesian theory is applied to the regional long-term inflowing multi-target joint probability prediction technique for the first time and different levels of the layered Bayesian model carry out MCMC sampling on the independent characteristic parameters and population characteristic parameters, so that the inflowing multi-target probability prediction model is perfected from perspectives of the physical mechanism and the algorithm and thus the accuracy of the regional long-term inflowing multi-target joint probability prediction is improved.