Ice-influenced hydrologic and hydrodynamic processes often cause floods in cold regions of the globe. These floods are typically associated with ice jams and can have negative socio-economic impacts, while their impacts on riverine ecosystems can be both detrimental and beneficial. Several methods have been proposed for constructing frequency distributions of ice-influenced annual peak stages where historical data are scarce, or for estimating future frequencies under different climate change scenarios. Such methods rely on historical discharge data, which are generally easier to obtain than peak stages. Future discharges can be simulated via hydrological models, driven by climate-model output. Binary sequences of historical flood/no-flood occurrences have been studied using logistic regression on physics-based explanatory variables or exclusively weather-controlled proxies, bypassing the hydrological modelling step in climate change projections. Herein, background material on relevant river ice processes is presented first, followed by descriptions of various proposed methods to quantify flood risk and assess their advantages and disadvantages. Discharge-based methods are more rigorous; however, projections of future flood risk can benefit from improved hydrological simulations of winter and spring discharges. The more convenient proxy-based regressions may not adequately reflect the controlling physics-based variables, while extrapolation of regression results to altered climatic conditions entails further uncertainty.