Based on using the Johnson-Cook dynamic constitutive model and shear failure critera, the machining distortion problem of asymmetric thin-walled part is studied from the angle of clamping position. The data of machining distortion and clamping position were acquired. The mathematical model between them was built by multivariate nonlinear regressing. Clamping position was optimized by genetic algorithm that has a strong global searching ability. The results show that the clamping position after optimizing can greatly reduce the machining distortion, and the average distortion is reduced 62% as compared with the traditional one (uniform distribution of clamping elements).