Abstract: Traditional optimization methods for the multi-objective problem are often transformed into a single objective optimization problem through some techniques. A certain degree of subjectivity due to these transformation processes needs to artificially define some parameters, and optimization results are usually more sensitive to the weighting coefficient. Additionally, because the objective function of multi-objective optimization problem is often nonlinear or discontinuous, using the traditional mathematical programming methods is relatively low efficiency. Especially in the field of automotive aerodynamics, when facing multi-objective optimization problems, a lot of work are required frequently and it is difficult to obtain ideal results. Therefore, a intelligent algorithm was adopted to solve multi-objective problem in this study. Combining CFD software and intelligent algorithm, the aerodynamic shape of SUV was optimized based on multi-objective optimization and design technique. The optimized object is a high-speed running SUV with a jar sunroof. The minimum aerodynamic drag, suitable aerodynamic lift (here is 0), and the minimum pressure on the trailing edge of the sunroof were taken as the optimization objective. The key shape parameters of the vehicle were selected as the design variables. The data mining technology was applied to evaluate the regulation and effects between design variables and three objective functions, and wind tunnel experiments were carried out to verify the validity of the algorithm method. Results show that through the non-dominated sorting genetic algorithm Ⅱ (NSAG-Ⅱ) method, under the premise of that other design objectives meet the design requirements, the aerodynamic drag of the optimized vehicle successfully decreases by 9.5%, and the accuracy of the intelligent algorithm is verified by wind tunnel test. The research indicates that the aerodynamic design of automobile aerodynamic shape based on an intelligent algorithm has excellent applicability and practical significance, and provides an efficient, accurate and reliable aerodynamic shape optimization method for the multi-objective optimization and design.