Abstract: Test data generation is a hard problem in software testing area and its implementation can greatly enhance the automation of software testing. The main contribution of this paper lies in that a pratical model, which utilizes genetic algorithms as searching policy to generate software structural test data , is proposed. To achieve higher performance, such issues as coding strategy, evaluation function construction and instrumentation are addressed in detail. In particular, much emphasis is put on algorithms operator evolution because the employ of operator is a key factor which can highly affect algorithms efficiency. Finally, some convincing experiment results are presented which show that our genetic algorithms-based method is more efficient comparing with other methods