Northern peatlands are one of the largest natural sources of atmospheric methane (CH 4), and it is important to understand the mechanisms of CH 4 loss from these peatlands so that future rates of CH 4 emission can be predicted. CH 4 is lost to the atmosphere from peatlands by diffusion, by plant transport, and as bubbles (ebullition). We argue that ebullition has not been accounted for properly in many previous studies, both in terms of measurement and the conceptualization of the mechanisms involved. We present a new conceptual model of bubble buildup and release that emphasizes the importance of near-surface peat as a source of atmospheric CH 4. We review two possible approaches to modeling bubble buildup and loss within peat soils: the recently proposed bubble threshold approach and a fully computational-fluid-dynamics approach. We suggest that neither satisfies the needs of peatland CH 4 models, and we propose a new reduced-complexity approach that conceptualizes bubble buildup and release as broadly similar to an upside down sandpile. Unlike the threshold approach, our model allows bubbles to accumulate at different depths within the peat profile according to peat structure, yet it retains the simplicity of many cellular (including cellular automata) models. Comparison of the results from one prototype of our model with data from a laboratory experiment suggests that the model captures some of the key dynamics of ebullition in that it reproduces well observed frequency-magnitude relationships. We outline ways in which the model may be further developed to improve its predictive capabilities.