The eddy covariance method was used to determine turbulent heat fluxes and carbon dioxide fluxes inside a boreal spruce – pine forest (2.5m above the forest floor) during the growing season in 1994 and 1995. Different data quality tests and spectral analysis were applied, confirming that most of the data collected inside the forest canopy, can be used to determine fluxes. Results of hourly averaged water- and carbon fluxes are compared to flux data measured continuously above the canopy.Large nonstationarities in sensible heat flux can be explained by nonlocal transport phenomena. Latent heat and carbon dioxide fluxes were more stationary because the sink/source strengths of water and carbon dioxide at the soil surface are more homogeneous compared to sources/sinks of sensible heat. Turbulent transport in the trunk space is caused by large intermittent eddies of 5–100m size, deduced from spectral analysis. Evaporation from soil and soil vegetation accounts for 10% of the total stand evaporation with rates between 0.1 and 0.6mm per day. In the daytime, the carbon loss from the soil is partly compensated by carbon uptake from the soil vegetation, resulting in flux rates of 0.45–0.9μmolm−2s−1. During the night, carbon fluxes of 0.1–3.6μmolm−2s−1(mean 2μmolm−2s−1) were observed under the canopy. Above the canopy, daily carbon uptake varied between 15 and 22μmolm−2s−1 near noon (daytime mean 9.5μmolm−2s−1).