Thermal microrefugia, sites within a landscape which are relatively protected from temperature extremes and warming trends, may be necessary for the conservation of animal species as climates warm. In freshwater environments, cold water fish species such as Pacific salmonids already rely on thermal microrefugia to persist in the southern extent of their range. Identifying sites that can provide such thermal microrefugia is essential to inform conservation and management decisions. At present, however, there is no consensus on appropriate ways to identify such sites, and multiple approaches are presented in the literature. Here, we use high resolution thermal data from two cold-water zones created at the confluence of tributaries with a warm main channel of the South Fork Eel River in Northern California to demonstrate that the characteristics of cold zones identified as thermal microrefugia are not robust to the choice of microrefugium definition. Common definitions disagree on the existence, volume, seasonal temporal trends, and diurnal variations in microrefugia at the two confluences. The disagreements arise from the complex interaction between the confluence mixing regime with diurnal/seasonal warming patterns and how this shapes the distribution of water temperature at the confluences, and how it is classified by each definition. The disagreements cannot be resolved by simple bias correction approaches. Given that all existing definitions rely on simplifications that negate the nuanced way fish use cold water zones and respond to thermal stress, alternative observation and classification approaches may be needed to characterize cold zones in rivers as thermal microrefugia.