The response of soil respiration to varying environmental factors was studied in four Picea abies stands (47-, 87-, 111- and 146-year old) during the 1998 growing season. While within-site variations of soil CO 2 efflux (up to 1.6 μmol CO 2 m −2 s −1) were larger than their diurnal variability (<0.25 μmol CO 2 m −2 s −1), spatial variations within a site were smaller than seasonal changes in soil respiration rates (up to 4.4 μmol CO 2 m −2 s −1). Highest within-site variability of soil efflux was generally found during the summer months when maximum flux rates of 4–6 μmol CO 2 m −2 s −1 were reached (coefficient of variation 40%). Soil temperatures (in the O f and O h layers, and A h horizon) showed a pronounced seasonal course, in contrast to soil moisture. An exponential equation best described the relationships between soil temperature in the O f layer and soil CO 2 efflux ( r 2 between 0.75 and 0.81). However, an Arrhenius type equation always resulted in lower r 2 values (0.52–0.71). The Q 10 values ranged between 2.39 (146-year old stand) and 3.22 (87-year old stand), averaging 2.72 for the P. abies stands within the watershed. The removal of litter and organic layers generally affected soil CO 2 efflux negatively. In three of the four P. abies stands (47-, 87-, 146-year old stands), soil respiration rates were reduced by 10–20% after removal of the L and O f layer, and by 30–40% after removal of the L and most of the O f and O h layers. Thus, mineral soil respiration seemed to contribute a major fraction to the total soil CO 2 flux (>60%). Trenching shallow fine roots during collar insertion and mechanical inhibition of root in-growth during the following months allowed fine root respiration to be separated from microbial respiration only in times of highest root growth. Microbial respiration seemed to dominate the respiratory CO 2 loss from the forest floor (>70%). The comparison of the annual soil CO 2 efflux in the 47-year old P. abies stand (about 710 g C m −2 yr −1) with annual litterfall and root net primary productivity estimates supported this conclusion.