Mass loss from fringing ice caps and glaciers in Greenland accounted for 20% of the ice sheet’s mass loss between 2003 and 2008 and could be responsible for up to 11 mm of sea- level rise by the end of the century. This study therefore applies a 3D, full-Stokes glaciological flow model, Elmer/Ice, to Flade Isblink, the largest fringing ice cap in Greenland, to investigate its basal conditions and the mechanism of a surge of two of its major outlet glaciers observed around the turn of the millennium. The results show that the ice cap is largely cold-based outside areas of fast flow, but that freezing also dominates in sliding areas. This implies an additional hydrological or heat input to allow the persistence of fast flow, the most likely candidate being surface meltwater and the associated latent heat release, leading to cryo-hydrologic warming. Model results indicate that the surge mechanism is a soft-bed thermal surge assisted by a hydro-thermodynamic feedback, with an estimated return period of 20-43 years. A till layer of 0.2-0.35 m thickness is inferred beneath sliding areas according to the hydrological budgets estimated for the surging glaciers. Results also indicate that this type of surge can self-terminate even without thinning of overlying ice, as slowdown will naturally reduce the frictional heat flux at the bed and thereby increase the release of latent heat of fusion due to basal freezing. Faster basal freezing will, in turn, produce a stronger bed and even slower ice flow due to withdrawal of pore water from the underlying till layer, which acts as a reservoir of latent heat as well as water. On the basis of these results, it is hypothesised that surging on Flade Isblink will increase in frequency over the coming decades, as more heat and meltwater will be stored in the subglacial till layer under global warming.