Due to the fact that the complexity of loads and uncertainties of random variables affect the reliability of defective casings, with consideration to the disadvantages of the deterministic approach, in this paper a probabilistic assessment method is employed based on previously established safety evaluation criteria for casings with corrosion defects in thermal recovery wells. In addition, Monte Carlo simulation is proposed to analyze the casing reliability under different remaining strengths. Sensitivity analysis is then performed to rank the influence of various variables for casing failure, and finally the influence law of the main parameters on the maximum Von Mises stress of defective casing is summarized.