Considering both material fatigue scatter and size effects， a probabilistic fatigue life analysis method for structures is developed based on the weakest link theory. The fatigue life distribution of the smooth specimens is modeled using both log-normal and Weibull distributions. This method is applied to predict and analyze the fatigue life and fatigue life scatter coefficient of TC11 titanium alloy simulated specimens and disk hub based on the fatigue life distribution of the TC11 smooth specimens. The results show that both models based on the weakest link theory accurately predict the median fatigue life of TC11 specimens， with predictions falling within twice the scatter range of experimental results. The log-normal model’s prediction of fatigue life scatter is closer to experimental data than that of the Weibull model. For the disk hub region， the life predictions of both models are similar at higher failure probabilities， and the predictions by the log-normal model are significantly larger than those by the Weibull distribution model when the failure probabilities are low. Furhter， the predictions on the fatigue life scatter of this region by the weakest link model based on the lognormal distribution are significantly smaller than the scatter coefficients of the smooth specimens and the simulated ones. The proposed log-noarml model can better analyze and predict the influence of size effects on structural fatigue scatter than the Weibull distribution model.