Embedded realtime systems are extensively used in safetycritical environments, such as transportation, aerospace  and nuclear power systems. Although system design may not have any defects, random failures due to wear of physical components or sudden changes in the environment can  cause system hazards during operation. Currently, the hazard analysis methods based on failure propagation model either only consider failure propagation time or just failure probability, and do not comprehensively analyze the impact of the failure propagation time and the failure probability on the hazard analysis. Timed failure propagation graphs (TFPGs) are usually used to model the failure propagation process in the design phase of a safetycritical system, which includes failure propagation delay modeling. Considering the effect of the uncertainty of failure propagation path on the probability of the hazard occurrence, we propose a hazard analysis method, which uses the probabilisticTFPGs to model the failure propagation process. We also design an analysis algorithm to obtain the correlation between occurrence time and occurrence probability. Finally, a case is given to demonstrate the feasibility of the proposed approach.