A robust model of CH4 emission from terrestrial ecosystems should be capable of simulating the temporal and spatial variability that characterizes field measurements. Such a model should couple a biologically based treatment of microbial CH4 transformations with a physically based treatment of heat, solute, and gas transfer vertically and laterally through soils. These processes are coupled in the ecosystem model Ecosys, which was tested against CH4 effluxes measured with surface chambers and a flux tower at a beaver pond in the BOREAS Northern Study Area. Spatial and temporal variation of CH4 effluxes in the model encompassed that measured by surface chambers and the flux tower. Both modeled and measured CH4 effluxes rose from 1.0 ¿mol m-2 s-1 and >0.5 g C m-2 d-1 at the pond margin. Larger effluxes occurred in the model when warming pond sediments generated episodic bubbling events. Annual CH4 effluxes in the model rose from <1 g C m-2 at sites above the pond to 76 g C m-2 at the pond margin. Annual totals included several brief but rapid efflux events during thawing and warming of soil and pond sediments that are frequently missed by surface measurements. Annual CH4 effluxes predicted after 100 years under an IS92a-driven climate change scenario rose by ~20% from the pond, but changed little from the surrounding landscape, indicating topographic variation in response of CH4 effluxes to climate change.