Software testing is an important part of software development, which can effectively improve software reliability and its quality. The reuse of test cases can improve the efficiency of testing and reduce the workload of software testing. Therefore, an approach to test case reuse between similar programs based on keyword flow graph is proposed. The method reuses test data generated by programs to their similar programs. Thus, the first step in exploring reuse of test cases is to determine similarities between programs. To determine the similarity of programs, a comparison method of similarity based on keyword flow graph is given. Firstly, the keywords in the program code are stored in the nodes corresponding to the flow graph to construct the keyword flow graph. Then, the maximum common sub-graph of keyword flow graph of the maximum programs tested is searched by using dynamic programming algorithm. Finally, the similarity of the programs is computed according to the common sub-graph distance algorithm. Programs with a high similarity can be used with test case reuse methods. In utilizing genetic algorithm to generate test cases, the test cases with high fitness of similar programs are selected. To speed up the efficiency of test case generation, the population intersects with these test cases continuously in the process of evolution. Experiments show that the reuse of test cases in test generation of similar programs has obvious advantages over traditional methods in terms of coverage and average evolution times.