Abstract: River ice flood occurs frequently in regions with high latitudes and high altitudes, often resulting in ice jams or dam floods that pose serious threats to personal and property safety. Therefore, comprehensive consideration of all factors to conduct scientific risk evaluation is the primary prerequisite and essential requirement for preventing the risk of freezing flood disasters. This study establishes an ice flood disaster risk evaluation model based on catastrophe theory, introduces the Gray Relational Analysis method to rank basic indicators and evaluates the ice flood disaster risk in the upper reaches of Heilongjiang River. Considering the correlation between different indicators, the Pearson correlation coefficient method was employed to streamline the indicator set and select representative years. The obtained results were compared based on the risk ranking of these representative years and Hierarchical Cluster Analysis-Catastrophe Theory. The findings reveal that under the influence of hydrological, meteorological, social and economic factors, the risk level of ice flood disaster in the three districts of Mohe, Tahe and Huma in the upper reaches of Heilongjiang River during the period from 2000 to 2020 has shown an overall trend of increasing and then decreasing. Among them, factors such as snow depth, water level, ice thickness, and temperature show a remarkably high correlation with the frequency of ice flood disasters. Social factors such as population, agriculture, economy, and healthcare also influence the risk level after ice flood events. The Mohe River section, due to its complex and steep river terrain, harsh climatic conditions, and relatively high population density, exhibits an overall elevated risk level. The risk membership values of the Gray Relational Analysis-Catastrophe Theory model were distributed in the range of 0.85 to 0.93, and it has better risk ranking and safety margin than the Hierarchical Cluster Analysis-Catastrophe Theory model, which verifies the validity and practicability of the model in the evaluation of the risk of ice flood disaster, and provides a new scientific method and theoretical basis for further research into ice flood disaster risk evaluation.