[Objective]Traditional rainfall stability analyses of tailing dams often fail to capture the randomness of actual rainfall processes. To address this issue, a fragility-curve-based method is proposed for safety assessment of tailing dams during the flood season, with the aim of probabilistically characterizing the stability risk of tailing dams under random rainfall conditions.[Methods]Within the framework of unsaturated seepage analysis and limit equilibrium theory, a bounded random cascade model was introduced to simulate random rainfall processes. The generated rainfall time series were then applied as rainfall boundary conditions in Geo-Studio for unsaturated seepage and stability analyses. Three stability limit states were defined according to relevant specifications and engineering experience. The exceedance probabilities of these limit states were calculated under different combinations of rainfall duration and cumulative rainfall, and the corresponding fragility curves for flood-season safety assessment of tailing dams were established.[Results]The obtained fragility curves exhibited typical S-shaped characteristics. As rainfall duration increased, a larger cumulative rainfall was required to reach the same exceedance probability. Under the same cumulative rainfall condition, the exceedance probability of each limit state decreased with increasing rainfall duration. The analysis of actual rainfall events further showed that the proposed method can quantify the probability of a tailing dams reaching different stability limit states during continuous rainfall.[Conclusion]The proposed method, which combines the bounded random cascade model with fragility curves, can effectively describe the influence of random rainfall processes on tailing dams stability and provide a probabilistic evaluation of dam safety under different rainfall scenarios. In addition, by incorporating the recovery time T_r, the method can be further applied to dynamic safety assessment of tailings dams under intermittent rainfall conditions.