Characterizing the use of carbon (C) reserves in trees is important for understanding regional and global C cycles, stress responses, asynchrony between photosynthetic activity and growth demand, and isotopic exchanges in studies of tree physiology and ecosystem C cycling. Using an inadvertent, whole-ecosystem radiocarbon ((14)C) release in a temperate deciduous oak forest and numerical modeling, we estimated that the mean age of stored C used to grow both leaf buds and new roots is 0.7 years and about 55% of new-root growth annually comes from stored C. Therefore, the calculated mean age of C used to grow new-root tissue is similar to 0.4 years. In short, new roots contain a lot of stored C but it is young in age. Additionally, the type of structure used to model stored C input is important. Model structures that did not include storage, or that assumed stored and new C mixed well (within root or shoot tissues) before being used for root growth, did not fit the data nearly as well as when a distinct storage pool was used. Consistent with these whole-ecosystem labeling results, the mean age of C in new-root tissues determined using 'bomb-(14)C' in three additional forest sites in North America and Europe (one deciduous, two coniferous) was less than 1-2 years. The effect of stored reserves on estimated ages of fine roots is unlikely to be large in most natural abundance isotope studies. However, models of root C dynamics should take stored reserves into account, particularly for pulse-labeling studies and fast-cycling roots (< 1 years).