Carbon allocation plays a critical role in forest ecosystem carbon cycling. We reviewed existing literature and compiled annual carbon budgets for forest ecosystems to test a series of hypotheses addressing the patterns, plasticity, and limits of three components of allocation: biomass, the amount of material present; flux, the flow of carbon to a component per unit time; and partitioning, the fraction of gross primary productivity (GPP) used by a component. Can annual carbon flux and partitioning be inferred from biomass? Our survey revealed that biomass was poorly related to carbon flux and to partitioning of photosynthetically derived carbon, and should not be used to infer either. Are component fluxes correlated? Carbon fluxes to foliage, wood, and belowground production and respiration all increased linearly with increasing GPP (a rising tide lifts all boats). Autotrophic respiration was strongly linked to production for foliage, wood and roots, and aboveground net primary productivity and total belowground carbon flux (TBCF) were positively correlated across a broad productivity gradient. How does carbon partitioning respond to variability in resources and environment? Within sites, partitioning to aboveground wood production and TBCF responded to changes in stand age and resource availability, but not to competition (tree density). Increasing resource supply and stand age, with one exception, resulted in increased partitioning to aboveground wood production and decreased partitioning to TBCF. Partitioning to foliage production was much less sensitive to changes in resources and environment. Overall, changes in partitioning within a site in response to resource supply and age were small (<15% of GPP), but much greater than those inferred from global relationships. Across all sites, foliage production plus respiration, and total autotrophic respiration appear to use relatively constant fractions of GPP – partitioning to both was conservative across a broad range of GPP – but values did vary across sites. Partitioning to aboveground wood production and to TBCF were the most variable – conditions that favored high GPP increased partitioning to aboveground wood production and decreased partitioning to TBCF. Do priorities exist for the products of photosynthesis? The available data do not support the concept of priorities for the products of photosynthesis, because increasing GPP increased all fluxes. All facets of carbon allocation are important to understanding carbon cycling in forest ecosystems. Terrestrial ecosystem models require information on partitioning, yet we found few studies that measured all components of the carbon budget to allow estimation of partitioning coefficients. Future studies that measure complete annual carbon budgets contribute the most to understanding carbon allocation.