This study presents a methodology that analyzes the fatigue reliability of a composite wind turbine blade considering wind load uncertainty. To facilitate the reliability analysis of wind turbine design, the turbulent random wind field has been simulated and characterized by two random variables, 10-minute mean wind speed and 10-minute turbulence intensity factor. The well-known Weibull distribution of 10-minute mean wind speed has been validated by statistically analyzing measured wind speed data. A log-logistic distribution is first proposed to represent the distribution of 10-minute turbulence intensity factor. By using both the mean wind speed and the turbulence intensity factor, the chaotic characteristic of a random wind field can be accurately rendered. The uncertainties of parameters determining the Weibull and log-logistic distribution are further studied such that the spatiotemporal wind uncertainty can be accurately represented. A hierarchical expanded wind uncertainty representation method is proposed for reliability analysis of wind turbine blades. A comprehensive procedure, including random wind simulation, aerodynamic analysis, composite structural analysis and fatigue damage calculation has been realized to predict the fatigue life of a simulated blade model. The reliability of a 5-MW reference wind turbine blade is evaluated to investigate the effect of the spatiotemporal wind uncertainty towards fatigue life.