We measured annual C storage, or net ecosystem production (NEP), from 1999 to 2003 in an aspen-dominated, mixed-deciduous forest in Michigan, USA. Measurements of the annual production of above- and below-ground live and dead mass, foliar herbivory, and soil respiration were used to develop biometric estimates of NEP (NEPB). Eddy-covariance measurements made above the canopy were used to construct meteorological estimates of NEP (NEPM). Over the 5-year study period, C stored annually in live mass (above- and below-ground wood) averaged 2.13 Mg C/ha/year, while annual detritus production (fine roots, leaf litter, woody debris) averaged 4.35 Mg C /ha/year. Fine root inputs were the largest component (41%) of annual net primary production. The forest was a consistent C sink, with annual NEP ranging from 0.80 to 1.98 Mg C ha1 year1, comparable in magnitude to other eastern North American forests. When annual NEPB and NEPM were compared in a given year, they differed from each other by 13v148%. However, when compared over 5 years, these independent estimates of C storage converged to within 1% of each other. Differences between same-year biometric and meteorological NEP estimates were explained in part by a lag between late-season net canopy photosynthesis and C allocated to tree growth the following spring. Weekly assessments of bole radial growth in 2001 and 2002 indicated that woody mass growth began prior to positive net canopy photosynthesis in the spring and that >25% of annual photosynthetic C assimilation occurred after growth had stopped in the autumn. Thus, while the temporal separation between photosynthesis and growth reduced agreement between annual NEPB and NEPM estimates, these metrics converged over several years. These results suggest that the allocation of recent photosynthate to storage carbohydrates, rather than to immediate growth, may obscure the relationship between shorter term, annual biometric and meteorological C storage estimates. Moreover, the 5-year convergence of biometric and meteorological NEP estimates serves as an important cross-validation, demonstrating that both approaches can yield accurate forest C storage assessments.