Active methane seepage occurs congruent with a high density of up to 1 km-wide and 35 m deep seafloor craters (>100 craters within 700 km 2 area) within lithified sedimentary rocks in the northern Barents Sea. the crater origin has been hypothesized to be related to rapid gas hydrate dissociation and methane release around 15-12 ka BP, but the geological setting that enabled and possibly controlled the formation of craters has not yet been addressed. to investigate the geological setting beneath the craters in detail, we acquired high-resolution 3D seismic data. The data reveals that craters occur within ~250-230 Myr old fault zones. Fault intersections and fault planes typically define the crater perimeters. Mapping the seismic stratigraphy and fault displacements beneath the craters we suggest that the craters are fault-bounded collapse structures. The fault pattern controlled the craters occurrences, size and geometry. We propose that this triassic fault system acted as a suite of methane migration conduits and was the prerequisite step for further seafloor deformations triggered by rapid gas hydrate dissociation some 15-12 ka BP. Similar processes leading to methane releases and fault bounded subsidence (crater-formation) may take place in areas where contemporary ice masses are retreating across faulted bedrocks with underlying shallow carbon reservoirs. Manifestations of past and present seabed methane discharge, such as seafloor pockmarks 1 , gas hydrate pingos 2 , craters 3 , and mounds 4,5 appear widely along Arctic continental margins. While pockmarks have received considerable attention from the fluid flow research community and petroleum industry for many decades 6-9 , km-wide seabed craters and associated mounds have garnered less scrutiny. Various studies report craters appearing on the seafloor on the Arctic continental shelf 3,5,10 , in the Scotia Sea 11 , at the Chatham Rise offshore New Zealand 12 , and on buried glacial surfaces in several areas of the Arctic 13,14. These reports may reveal that such features are more widespread than previously thought. As opposed to the crater site at the Scotia Sea and Chatham Rise, the crater area in the Barents Sea emits methane gas from the seabed today. Andreassen, et al. 5 , and Long, et al. 15 , document spatial relationships between craters and gas expulsions, and hypothesize that the craters are collapse features caused by blow-outs of over-pressured methane accumulations. A causal relationship between methane dynamics and crater development, however, remains unclear. Methane seepage from the seabed is the largest source of methane to the ocean 16. Nevertheless, due to methane dissolution and subsequent aerobic oxidation in the water column e.g. 17-20 , methane from deep water seep sources contributes less to the atmospheric carbon budget than previously thought. Because the lifetime of a methane bubble in the ocean depends on how much methane is already dissolved in the water and for how long the bubble is exposed, shallow-water seeps and larger plum...