Fluxes of carbon dioxide, sensible and latent heat from a 55-year-old Douglas-fir stand and a recently (2000) clear-cut site on the east coast of Vancouver Island, Canada were measured using the eddy-covariance technique. The two research sites were located within 3 km of one another, experienced the same weather and had similar site characteristics. Over a year, the net uptake of atmospheric carbon by the mature forest, per unit surface area, was less than the net loss of carbon from the clear-cut. Between August 1, 2000 and July 31, 2001, net ecosystem productivity (NEP) was 3.3 t C ha^-1 y^-1 for the mature forest and -6.9 t C ha^-1 y^-1 for the clear-cut. These NEP estimates were corrected by replacing low wind speed night-time carbon dioxide fluxes with values obtained from the relationship between high wind speed night-time fluxes and temperature. At the mature forest, the combination of a high leaf area index of 6.7 m² m^-2 and the mild coastal climate resulted in daytime carbon uptake throughout the year. The greatest 24-h net carbon uptake rates occurred during May when respiration remained relatively low compared to gross ecosystem production due to cool temperatures and abundant light. In the clear-cut, the growth of the planted 2-year-old Douglas-fir seedlings and a sparse canopy of weed species through the spring and summer of 2001 resulted in a maximum leaf area index of 0.87 m² m^-2. Consequently, 24-h NEP approached zero during July 2001. It also approached zero during December and January when soil respiration was inhibited by near-freezing conditions. During the rest of the year, the clear-cut was a source of carbon at all times of the day with values of NEP generally between -1 and -3 g C m^-2 d^-1. The impact of the microclimate and surface energy fluxes on carbon dioxide exchange at the two sites is also discussed.