The epilimnion is the surface layer of a lake typically characterised as well mixed and is decoupled from the metalimnion due to a steep change in density. The concept of the epilimnion (and, more widely, the three-layered structure of a stratified lake) is fundamental in limnology, and calculating the depth of the epilimnion is essential to understanding many physical and ecological lake processes. Despite the ubiquity of the term, however, there is no objective or generic approach for defining the epilimnion, and a diverse number of approaches prevail in the literature. Given the increasing availability of water temperature and density profile data from lakes with a high spatio-temporal resolution, automated calculations, using such data, are particularly common, and they have vast potential for use with evolving long-term globally measured and modelled datasets. However, multi-site and multi-year studies, including those related to future climate impacts, require robust and automated algorithms for epilimnion depth estimation. In this study, we undertook a comprehensive comparison of commonly used epilimnion depth estimation methods, using a combined 17-year dataset, with over 4700 daily temperature profiles from two European lakes. Overall, we found a very large degree of variability in the estimated epilimnion depth across all methods and thresholds investigated and for both lakes. These differences, manifesting over high-frequency data, led to fundamentally different understandings of the epilimnion depth. In addition, estimations of the epilimnion depth were highly sensitive to small changes in the threshold value, complex thermal water column structures, and vertical data resolution. These results call into question the custom of arbitrary method selection and the potential problems this may cause for studies interested in estimating the ecological processes occurring within the epilimnion, multi-lake comparisons, or long-term time series analysis. We also identified important systematic differences between methods, which demonstrated how and why methods diverged. These results may provide rationale for future studies to select an appropriate epilimnion definition in light of their particular purpose and with awareness of the limitations of individual methods. While there is no prescribed rationale for selecting a particular method, the method which defined the epilimnion depth as the shallowest depth, where the density was 0.1 kg m−3 more than the surface density, may be particularly useful as a generic method.