This study compares the CH4 fluxes from two arctic wetland sites of different annual temperatures during 2004 to 2006. The PEATLAND-VU model was used to simulate the emissions. The CH4 module of PEATLAND-VU is based on the Walter-Heimann model. The first site is located in northeast Siberia, Indigirka lowlands, Kytalyk reserve (70 degrees N, 147 degrees E) in a continuous permafrost region with mean annual temperatures of -14.3 degrees C. The other site is Stordalen mire in the eastern part of Lake Tornetrask (68 degrees N, 19 degrees E) ten kilometres east of Abisko, northern Sweden. It is located in a discontinuous permafrost region. Stordalen has a sub arctic climate with a mean annual temperature of -0.7 degrees C. Model input consisted of observed temperature, precipitation and snow cover data. In all cases, modelled CH4 emissions show a direct correlation between variations in water table and soil temperature variations. The differences in CH4 emissions between the two sites are caused by different climate, hydrology, soil physical properties, vegetation type and NPP. For Kytalyk the simulated CH4 fluxes show similar trends during the growing season, having average values for 2004 to 2006 between 1.29-2.09 mg CH4 m(-2) hr(-1). At Stordalen the simulated fluxes show a slightly lower average value for the same years (3.52 mg CH4 m(-2) hr(-1)) than the observed 4.7 mg CH4 m(-2) hr(-1). The effect of the longer growing season at Stordalen is simulated correctly. Our study shows that modelling of arctic CH4 fluxes is improved by adding a relatively simple hydrological model that simulates the water table position from generic weather data. Our results support the generalization in literature that CH4 fluxes in northern wetland are regulated more tightly by water table than temperature. Furthermore, parameter uncertainty at site level in wetland CH4 process models is an important factor in large scale modelling of CH4 fluxes.