Anoxic wetlands are an important source for the greenhouse gas CH4, much of which is emitted in form of gas bubbles. The conditions for formation of gas bubbles have recently been described by an analytical model, which allows the prediction of fluxes by first physical principles using the knowledge of gas concentration profiles and/or gas production rates. We tested parts of this model by experiments using microcosms of flooded, non-vegetated and homogeneous rice field soil incubated under different gas atmospheres and at different temperatures. In these experiments we determined rates of CH4 and CO2 production, upper boundaries of the bubble zone, gas-filled porosities and vertical profiles of dissolved CH4, CO2 and N2. The results of our experiments confirmed that by knowing only one of the following parameters, i.e. CH4 production, diffusive CH4 flux and depth of upper boundary of bubble zone, the remainder could be predicted from the model. On average, predicted values differed from experimental ones by a factor of 0.4 –2.7, depending on which parameter was taken as an input for the model. It was possible to predict the percentage of gas bubble flux from measured CH4 emission rates under the experimental conditions, which was on the order of 90%. The confrontation of the model with experimental data showed that the effect of the shallow upper oxic layer on bubble formation was negligible and that the CH4 diffusive flux is easily underestimated by experiments lacking sufficient spatial resolution. Therefore, CH4 production rates lower than in our microcosms would allow a more precise test of the model by creating less steep concentration gradients, which, however, would require long incubation times to purge the dissolved N2 from the soil by ebullition and to reach true steady state.