In order to obtain accurate probability integration method(PIM) parameters for surface movement of multi-panel mining, a genetic algorithm(GA) was used to optimize the parameters. As the measured surface movement is affected by more than one mining panel, traditional PIM parameter inversion model is difficult to ensure the reliability of the results due to the complexity of rock movement. With crossover,mutation and selection operators, GA can perform a global optimization search and has high computation efficiency. Compared with the pattern search algorithm, the fitness function can avoid falling into local minima traps. GA reduces the risk of local minima traps which improves the accuracy and reliability with the mutation mechanism. Application at Xuehu colliery shows that GA can be used to inverse the PIM parameters for multi-panel surface movement observation, and reliable results can be obtained. The research provides a new way for back-analysis of PIM parameters for mining subsidence under complex conditions. In order to obtain accurate probability integration method (PIM) parameters for surface movement of multi-panel mining, a genetic algorithm (GA) was used to optimize the parameters. As the measured surface movement is affected by more than one mining panel, traditional PIM Compared with the pattern optimization algorithm, the fitness of the results due to the complexity of rock movement function can avoid falling risk to local minima traps. GA reduces the risk of local minima traps which improves the accuracy and reliability with the mutation mechanism. Application at Xuehu colliery that that GA can be used to inverse the PIM parameters for multi-panel surface movement observation , and reliable results can be obtained. The research provides a new way for back-analysis of PIM parameters for mining subsidies ence under complex conditions.