Seasonal changes in the water and energy exchanges over a pine forest in eastern Siberia were investigated and compared with published data from a nearby larch forest. Continuous observations (April to August 2000) were made of the eddy-correlation sensible heat flux and latent heat flux above the canopy. The energy balance was almost closed, although the sum of the turbulent fluxes sometimes exceeded the available energy flux (R n G) when the latent heat flux was large; this was related to the wind direction. We examined the seasonal variation in energy balance components at this site. The seasonal variation and magnitude of the sensible heat flux (H was similar to that of the latent heat flux ( E), with maximum values occurring in mid-June. Consequently, the Bowen ratio was around 1Γ0 on many days during the study period. On some clear days just after rainfall, E was very large and the sum of H and E exceeded R n G. The evapotranspiration rate above the dry canopy from May to August was 2Γ2 mm day 1 . The contributions of understory evapotranspiration (E u ) and overstory transpiration (E o ) to the evapotranspiration of the entire ecosystem (E t ) were both from 25 to 50% throughout the period analysed. These results suggest that E u plays a very important role in the water cycle at this site. From snowmelt through the tree growth season (23 April to 19 August 2000), the total incoming water, comprised of the sum of precipitation and the water equivalent of the snow at the beginning of the melt season, was 228 mm. Total evapotranspiration from the forest, including interception loss and evaporation from the soil when the canopy was wet, was 208-254 mm. The difference between the incoming and outgoing amounts in the water balance was from C20 to 26 mm. The water and energy exchanges of the pine and larch forest differed in that E and H increased slowly in the pine forest, whereas E increased rapidly in the larch forest and H decreased sharply after the melting season. Consequently, the shape of the Bowen ratio curves at the two sites differed over the period analysed, as a result of the differences in the species in each forest and in soil thawing.