James Bay Lowland peatlands are environments with unique hydrologic characteristics that challenge some basic assumptions embedded within many hydrology models, including topographically-driven lateral flows and hydrologic connectivity of all terrestrial landscape elements within the stream network. With increasing resource development in northern lowland regions of Canada, more rigorous and honest appraisal of modelling capabilities and deficiencies is warranted. This study was initiated with the following two objectives: (1) to compare the performance of two popular conceptual
rainfall-runoff models, TOPMODEL and HBV, for rainfall-runoff simulation in a James Bay Lowland peatland complex in the James Bay Lowlands, and (2) to compare regionalization methods to maximize the predictive value of available landscape information to improve model calibration using HBV. HBV was found to outperform TOPMODEL, which was altogether unsuitable for this environment. Regionalization analyses and results favoured empirical methods such as artificial neural networks to improve predictive capabilities of the HBV model.