Abstract: The test data generated by random mutation in fuzz testing destroys the input specification of the target program, which leads to the failure of test data verification and low code coverage. Aiming at this problem, a fuzzing test data generation method was proposed based on dynamic construction of mutation strategy. The method was designed to use the feedback information of instrumentation to dynamically construct the control mutation strategy and the keyword mutation strategy, and to guide the fuzzer to generate test data with high coverage. Experimental results show that compared with random mutation, this method can improve the code branch coverage by about 40% on average. This method can effectively improve the efficiency of fuzz testing, and has a strong practical value.