The active season of tussock tundra was extended during two growing seasons (1995 and 1996) by snow removal in early season and prevention of snow accumulation in late season to test the effects of a longer growing season on tundra carbon exchange. Three treatments were established: extended season, extended season+soil warming, and controls. Soil warming was accomplished using cold-frame, resistance heating wire installed the year prior to the initiation of treatments. Diurnal courses of CO2 exchange were measured weekly using infrared gas analysis with enclosed chamber techniques. Methane fluxes were measured two to three times a season also using enclosure methods. In 1995, snowmelt occurred unusually early, and snow removal treatments increased the season only 9--10 days. In 1996 the early season was increased approximately 24 days. As expected, thaw depth, soil temperature, and plant growth were greater earlier in the extended season and extended season+soil heating plots. Methane fluxes in both seasons were low but tended to be higher in the extended season and soil heated plots. Net ecosystem CO2 fluxes were similar among treatments early in the season, with a tendency toward more positive fluxes (system loss) for the snow removal and warmed plots, possibly due to higher belowground respiration. During midseason, fluxes were similar among the treatments. Later in the season, fluxes of extended season and warmed plots tended to be lower (less carbon loss) than controls, especially in 1995. Totaled over the season, however, the fluxes of the three treatments did not statistically differ and represented losses to the atmosphere. Measurements of dark respiration in 1996 indicate that both respiration and uptake were increased on the extended season plots, resulting in similar net fluxes to controls. ¿ 1998 American Geophysical Union