Complicated software often contains many paths, and there is few effective method of generating test data to
cover these paths up to present. Therefore, a method of evolutionary generation of test data for many paths coverage based
on adaptive grouping is presented. During the process of evolution, the groups satisfying given conditions are merged
based on the similarity. Then the problem of generating test data is transformed into multi-objective optimization problems
with constraints whose number decreases gradually. A multi-population genetic algorithm is employed to solve the above
problems, especially, the strategy of forming new populations after merging some groups is presented. The proposed method
is applied to one benchmark program, and the experimental results show that, the method can decrease the time spent in
generating test data greatly, and provides a feasible approach to improve the efficiency of software testing.