Abstract Bottom-simulating reflections (BSRs) are probably the most commonly used indicators for gas hydrates in marine sediments. It is now widely accepted that BSRs are primarily caused by free gas beneath gas-hydrate-bearing sediments. However, our insight into BSR formation to date is mostly limited to theoretical studies. Two endmember processes have been suggested to supply free gas for BSR formation: (i) dissociation of gas hydrates and (ii) migration of methane from below. During a recent campaign of the German Research Vessel Sonne off the shore of Peru, we detected BSRs at locations undergoing both tectonic subsidence and non-sedimentation or seafloor erosion. Tectonic subsidence (and additionally perhaps seafloor erosion) causes the base of gas hydrate stability to migrate downward with respect to gas-hydrate-bearing sediments. This process rules out dissociation of gas hydrates as a source of free gas for BSRs at these locations. Instead, free gas at BSRs is predicted to be absorbed into the gas hydrate stability zone. BSRs appear to be confined to locations where the subsurface structure suggests focusing of fluid flow. We investigated the seafloor at one of these locations with a TV sled and observed fields of rounded boulders and slab-like rocks, which we interpreted as authigenic carbonates. Authigenic carbonates are precipitations typically found at cold vents with methane expulsion. We retrieved a small carbonate-cemented sediment sample from the seafloor above a BSR about 20 km away. This supported our interpretation that the observed slabs and boulders were carbonates. All these observations suggest that BSRs in Lima Basin are maintained predominantly by gas that is supplied from below, demonstrating that this endmember process for BSR formation exists in nature. Results from Ocean Drilling Program Leg 112 showed that methane for gas hydrate formation on the Peru lower slope and the methane in hydrocarbon gases on the upper slope is mostly of biogenic origin. The δ 13C composition of the recovered carbonate cement was consistent with biologic methane production below the seafloor (although possibly above the BSR). We speculate that the gas for BSR formation in Lima Basin also is mainly biogenic methane. This would suggest the biologic productivity beneath the gas hydrate zone in Lima Basin to be relatively high in order to supply enough methane to maintain BSRs.