Mutative Scale Immune Clonal Selection Algorithm (MSICSA) based on Multi-population is proposed. In the algorithm, the dominant position of global optimal solution was highlighted by the nonlinear scale transformation of objective function. Memory sub-population was extracted to exchange information between populations. Antibody population was divided into elite, normal and inferior sub-population. To enhance local and global search capabilities of MSICSA, adaptive Gaussian and uniform mutation were applied to elite and normal sub-population respectively and the inferior antibody was extinguished and replaced by new ones. By introducing the niche technology to increase the diversity of population distribution, the algorithm can prevent premature. Test functions and a space antenna optimization were tested. The results show that the optimization capability of MSICSA is more advanced than CLONALG and SGA, and the computational complexity is reduced.