Aiming at small sample condition of launch vehicle, this paper combined the sequential posterior odd test (SPOT) and Bayesian recursive estimation method, and it improves the traditional test sample size design from two aspects of hypothesis test and parameter estimation. When evaluating the launch reliability of launch vehicle, the proposed method comprehensively considers two types of risks and confidence requirements, makes up for the deficiency of prior information or field data, and formulates a reasonable and effective evaluation scheme, so as to make the test results more accurate and the test sample size less. Through example analysis, it is found that the proposed method can effectively reduce the demand for sample size, and the estimation results are true and reliable. It can be well used for launch reliability evaluation test of launch vehicle under small sample conditions.