Abstract: The purpose of test suit reduction is to improve the efficiency of software testing and reduce the cost of test. At the same time, the coverage of the original test cases should be ensured invariable and the number of test cases could be reduced as possible with the use of a certain strategy. Vector similarity algorithm is that the binary vector similarity function should be used firstly, then the similarity, coverage and redundancy test cases needed be calculated, and the best test cases is selected according to the calculation result, finally we will get the optimal or approximate optimal test cases after the reduction. Compared with the existing algorithm of GRE, experiments show that the number of test cases decreased by about 10% averagely and the whole redundancy rate of test cases decreased by about 13% averagely according to the vector similarity algorithm.