Composite laminated plates have been widely applied to aircraft primary structures. However, composites have remarkable mechanical anisotropy, and material design on stacking sequences is needed. In recent years genetic algorithms have been worthy of notice as solutions for combinatorial optimum problems. Some reserachers have applied the genetic algorithm for optimization of stacking sequence, but not for stacking number. Therefore, three new approaches were proposed and examined in detail in this study. A 'two species GA model' uses two species which differ in their stacking numbers. The choromosomes of meta populations have one less genes by one than the chromosomes of actual populations in this model. However, these populations are independent of each other. A 'mixed two species model' is similar to the two species GA except for crossover, and the two species exchange their genes with each other. A 'mutant model' uses only mutation. Using these three models, a bending stiffness optimization problem was examined. As a result, both the 'mixed two species model' and the 'mutant model' are applicable in the special case that there is little difference between local minimum points, such as in this problem.