Abstract.The European Alps, cradle of pioneer glacial studies, are one of the regions where geological markers of past glaciations are most abundant and well-studied. Such conditions make the region ideal for testing numerical glacier models based on simplified ice flow physics against field-based reconstructions, and vice-versa.Here, we use the Parallel Ice Sheet Model (PISM) to model the entire last glacial cycle (120-0 ka) in the Alps, using 5 horizontal resolutions of 2 and 1 km and up to 576 processors. Climate forcing is derived using present-day climate data from WorldClim and the ERA-Interim reanalysis, and time-dependent temperature offsets from multiple palaeo-climate proxies, among which only the EPICA ice core record yields glaciation during marine oxygen isotope stages 4 (69-62 ka) and 2 (34-18 ka) spatially and temporally consistent with the geological reconstructions, while the other records used result in excessive early glacial cycle ice cover and a late Last Glacial Maximum. Despite the low variability of this Antarctic-based climate 10 forcing, our simulation depicts a highly dynamic ice sheet, showing that Alpine glaciers may have advanced many times over the foreland during the last glacial cycle. Ice flow patterns during peak glaciation are largely governed by subglacial topography but include occasional transfluences and self-sustained ice domes. Modelled maximum ice surface is 861 m higher than observed trimline elevations in the upper Rhone Valley, yet our simulation predicts little erosion at high elevation due to cold-based ice. Finally, the Last Glacial Maximum advance, often considered synchronous, is here modelled as a time-15 transgressive event, with some glacier lobes reaching their maximum as early as 27 ka, and some as late as 21 ka.