AbstractCold climate rivers can experience avulsions due to both open water and ice jam flooding; however, most existing models for evaluating avulsions only consider open water flows. We present a novel approach for determining site‐specific probabilities of avulsion cause (open water vs. ice jam flooding) by combining historical flow data, channel cross‐sections and known avulsion history for the study area. The approach is applied to the Cumberland Marshes region of the Saskatchewan River in Central Canada, which experienced an avulsion in the mid‐1870s that some researchers have suggested may have been triggered by an ice jam. For the study area, overbank flooding was found to occur much more frequently due to ice jams than open‐water floods. Based on an average avulsion return period of 660 years in the study area, the probability of historical avulsions being caused by ice jam flooding was estimated to range from 61% and 80%, using a range of annual ice jam probabilities between 0.1 and 0.5. Results from the study suggest ice jam flooding as the most likely cause of the 1870s avulsion, which is also supported by historical evidence. The developed methodology is relatively simple to apply and could be easily implemented at other cold‐climate sites to evaluate avulsion risks.