The design of test-cases is one of the most important parts of software testing, which play an important role in guiding the post-testing and also is the fundamental guarantee of quality software. For the shortcoming of method raised by Moheb R. Girgis, an improved genetic algorithm for the automatic generation of data flow test-cases was proposed by introducing the branch functions and adaptive genetic strategies. Experiments show that the improved algorithm has a more increase in the performance of convergence rate and coverage rate.