As to the random sonic fatigue problem of thin walled structure of aerospace flight vehicle, estimation method for fatigue life based on Von Mises stress peak probability density is investigated. Assuming that Von Mises stress process satisfies three-parameter Weibull distribution, the peak probability density function of Von Mises stress is derived through the threshold crossing analysis and the peak distribution analysis of stationary random process. According to the Miner linear cumulative damage theory, the method for estimating the fatigue life in the frequency domain is established with the Von Mises stress peak probability density function applied. As an example, an aero-engine Combustor liner is considered, using coupled Finite Element Method (FEM) and Boundary Element Method (BEM) method, the structure vibration response to limited bandwidth Gaussian white noise is calculated. Based on the results, fatigue life of the structure is estimated by using the proposed method. Further more, the influences of the probability density function which is characterized by three-parameter Weibull distribution and two-parameter Weibull distribution respectivily for Von Mises stress response of the Combustor liner structure subjected to random accoustic loadings are discussed.