In the high Arctic valley of Adventdalen, Svalbard, sub-permafrost groundwater feeds several pingo springs distributed along the valley axis. The driving mechanism for groundwater discharge and associated pingo formation is enigmatic because wet-based glaciers are not present in the adjacent highlands and the presence of continuous permafrost seems to preclude recharge of the sub-permafrost groundwater system by either a subglacial source or a precipitation surplus. Since the pingo springs enable methane that has accumulated underneath the permafrost to escape directly to the atmosphere, our limited understanding of the groundwater system brings significant uncertainty to predictions of how methane emissions will respond to changing climate. We address this problem with a new conceptual model for open-system pingo formation wherein pingo growth is sustained by sub-permafrost pressure effects, as related to the expansion of water upon freezing, during millennial-scale basal permafrost aggradation. We test the viability of this mechanism for generating groundwater flow with decoupled heat (one-dimensional transient) and groundwater (three-dimensional steady state) transport modelling experiments. Our results suggest that the conceptual model represents a feasible mechanism for the formation of open-system pingos in lower Adventdalen and elsewhere. We also explore the potential for additional pressurisation and find that methane production and methane clathrate formation and dissolution deserve particular attention on account of their likely effects upon the hydraulic pressure. Our model simulations also suggest that the generally low-permeability hydrogeological units cause groundwater residence times to exceed the duration of the Holocene. The likelihood of such pre-Holocene groundwater ages is supported by the geochemistry of the pingo springs which demonstrates an unexpected seaward freshening of groundwater potentially caused by a palaeo-subglacial meltwater "wedge"from the Weichselian. Whereas permafrost thickness (and age) progressively increases inland, accordingly, the sub-permafrost meltwater wedge thins, and less unfrozen freshwater is available for mixing. Our observations imply that millennial-scale permafrost aggradation deserves more attention as a possible driver of sustained flow of sub-permafrost groundwater and methane to the surface because, although the hydrological system in Adventdalen at first appears unusual, it is likely that similar systems have developed in other uplifted valleys throughout the Arctic.