This doctoral dissertation presents a study of the glacial history of the North American Cordillera using numerical ice sheet modelling calibrated against field evidence. This area, characterized by the steep topography of several mountain ranges separated by large inter-montane depressions, was once covered by a large-scale ice mass: the former Cordilleran ice sheet. Because of the irregular topography on which the ice sheet formed, geological studies have often had only local or regional relevance, thus leaving the Cordilleran ice sheet least understood among Pleistocene ice sheets in terms of its extent, volume, and dynamics. Here, I present numerical simulations that allow quantitative reconstructions of the former ice sheet evolution based on approximated physics of glacier flow. These simulations show that the geometry of the Last Glacial Maximum Cordilleran ice sheet was largely controlled by sharp contrasts in regional temperature, precipitation, and daily temperature variability associated with the presence of mountain ranges. However, this maximum stage appears short-lived and out of balance with contemporaneous climate. During most of the simulated last glacial cycle, the North American Cordillera is characterized by an intermediate state of glaciation including isolated glaciers and ice caps covering major mountain ranges, the largest of which is located over the Skeena Mountains. The numerically modelled Cordilleran ice sheet appears in constant imbalance with evolving climate conditions, while the complexity of this transient response transcends that encapsulated in two-dimensional, conceptual models of ice sheet growth and decay. This thesis demonstrates the potential of numerical ice sheet modelling to inform on ice sheet history and former climate conditions over a glacial cycle, given that ice sheet models can be calibrated against field constraints. / <p>At the time of the doctoral defense, the following paper was unpublished and had a status as follows: Paper 4: Mansucript.</p>