A physically-based multi-layer snow model Snow-Atmosphere-Soil-Transfer scheme (SAST) and a land surface model Biosphere-Atmosphere Transfer Scheme (BATS) were employed to investigate how boreal forests influence snow accumulation and ablation under the canopy. Mass balance and energetics of snow beneath a Scots pine canopy in Finland at different stages of the 2003-2004 and 2004-2005 snow seasons are analyzed. For the fairly dense Scots pine forest, drop-off of the canopy-intercepted snow contributes, in some cases, twice as much to the underlying snowpack as the direct throughfall of snow. During early winter snow melting, downward turbulent sensible and condensation heat fluxes play a dominant role together with downward net longwave radiation. In the final stage of snow ablation in middle spring, downward net all-wave radiation dominates the snow melting. Although the downward sensible heat flux is comparable to the net solar radiation during this period, evaporative cooling of the melting snow surface makes the turbulent heat flux weaker than net radiation. Sensitivities of snow processes to leaf area index (LAI) indicate that a denser canopy speeds up early winter snowmelt, but also suppresses melting later in the snow season. Higher LAI increases the interception of snowfall, therefore reduces snow accumulation under the canopy during the snow season; this effect and the enhancement of downward longwave radiation by denser foliage outweighs the increased attenuation of solar radiation, resulting in earlier snow ablation under a denser canopy. The difference in sensitivities to LAI in two snow seasons implies that the impact of canopy density on the underlying snowpack is modulated by interannual variations of climate regimes.