Under complex environment and load, the deterioration of bridges resistance overtime will reduce the reliability and finally an unexpected failure occur suddenly impacting the durability, adaptability and safety of bridges. Therefore, it is of great significance to evaluate accurately the safety state of bridge and put forward timely reasonable suggestions and measures in maintenance, repair and resilience for bridge management and maintenance department. A Gamma stochastic process model for resistance degradation and a Poisson stochastic process model for live load were firstly established. Based on the two models and basic assumption that resistance and load effect are independent of each other, an expression for calculating structural failure probability was thus derived by using conditional probability formula, and an improved method to compute time-dependent probability calculation with Monte Carlo sampling was proposed. Finally, the Matlab language was used to compile the corresponding numerical program. As a benchmark example, the proposed method was used to investigate the attenuation law of reliability in resistance of a 20 years reinforced concrete T-beam bridge. Compared to the conventional method for calculating time-dependent probability , the present method not only is feasible and effective, but also has higher efficiency. In addition, taking the bridge resistance degradation model under corrosion mechanism as an example, The failure probability and reliability index of the bridge in the next 30 years are predicted.