The respiratory activities of plant roots, of their mycorrhizal fungi and of the free-living microbial heterotrophs (decomposers) in soils are significant components of the global carbon balance, but their relative contributions remain uncertain. To separate mycorrhizal root respiration from heterotrophic respiration in aboreal pine forest, we conducted a large-scale tree-girdling experiment, comprising 9 plots each containing about 120 trees. Tree-girdling involves stripping the stem bark to the depth of the current xylem at breast height terminating the supply of current photosynthates to roots and their mycorrhizal fungi without physically disturbing the delicate root-microbe-soil system. Here we report that girdling reduced soil respiration within 1-2 months by about 54% relative to respiration on ungirdled control plots, and that decreases of up to 37% were detected within 5 days. These values clearly show that the flux of current assimilates to roots is a key driver of soil respiration; they are conservative estimates of root respiration, however, because girdling increased the use of starch reserves in the roots. Our results indicate that models of soil respiration should incorporate measures of photosynthesis and of seasonal patterns of photosynthate allocation to roots.