Fatigue analysis of engine turbine blade is an essential issue. Due to various uncertainties during the manufacture and operation, the fatigue damage and life of turbine blade present randomness. In this study, the randomness of structural parameters, working condition and vibration environment are considered for fatigue life predication and reliability assessment. First, the low-cycle fatigue problem is modelled as stochastic static system with random parameters, while the high-cycle fatigue problem is considered as stochastic dynamic system under random excitations. Then, to deal with the two failure modes, the novel Direct Probability Integral Method (DPIM) is proposed, which is efficient and accurate for solving stochastic static and dynamic systems. The probability density functions of accumulated damage and fatigue life of turbine blade for low-cycle and high-cycle fatigue problems are achieved, respectively. Furthermore, the time–frequency hybrid method is advanced to enhance the computational efficiency for governing equation of system. Finally, the results of typical examples demonstrate high accuracy and efficiency of the proposed method by comparison with Monte Carlo simulation and other methods. It is indicated that the DPIM is a unified method for predication of random fatigue life for low-cycle and high-cycle fatigue problems. The rotational speed, density, fatigue strength coefficient, and fatigue plasticity index have a high sensitivity to fatigue reliability of engine turbine blade.