The terrestrial biosphere plays an important role in the global carbon cycle. In the 1994 Intergovernmental Panel Assessment on Climate Change (IPCC), an effort was made to improve the quantification of terrestrial exchanges and potential feedbacks from climate, changing CO sub(2), and other factors; this paper presents the key results from that assessment, together with expanded discussion. The carbon cycle is the fluxes of carbon among four main reservoirs: fossil carbon, the atmosphere, the oceans, and the terrestrial biosphere. Emissions of fossil carbon during the 1980s averaged 5.5 Gt y super(-1). During the same period, the atmosphere gained 3.2 Gt C y super(-1), and the oceans are believed to have absorbed 2.0 Gt C y super(-1). The regrowing forests of the Northern Hemisphere may have absorbed 0.5 Gt C y super(-1) during this period. Meanwhile, tropical deforestation is thought to have released an average 1.6 Gt C y super(-1) over the 1980s. While the fluxes among the four pools should balance, the average 1980s values lead to a 'missing sink' of 1.4 Gt C y super(-1). Several processes, including forest regrowth, CO sub(2) fertilization of plant growth (c. 1.0 Gt C y super(-1)), N deposition (c. 0.6 Gt C y super(-1)), and their interactions, may account for the budget imbalance. However, it remains difficult to quantify the influences of these separate but interactive processes. Uncertainties in the individual numbers are large, and are themselves poorly quantified. This paper presents detail beyond the IPCC assessment on procedures used to approximate the flux uncertainties. Lack of knowledge about positive and negative feedbacks from the biosphere is a major limiting factor to credible simulations of future atmospheric CO sub(2) concentrations. Analyses of the atmospheric gradients of CO sub(2) and super(13)CO sub(2) concentrations provide increasingly strong evidence for terrestrial sinks, potentially distributed between Northern Hemisphere and tropical regions, but conclusive detection in direct biomass and soil measurements remains elusive. Current regional-to-global terrestrial ecosystem models with coupled carbon and nitrogen cycles represent the effects of CO sub(2) fertilization differently, but all suggest long-term responses to CO sub(2) that are substantially smaller than potential leaf- or laboratory whole plant-level responses. Analyses of emissions and biogeochemical fluxes consistent with eventual stabilization of atmospheric CO sub(2) concentrations are sensitive to the way in which biospheric feedbacks are modeled by c. 15%. Decisions about land use can have effects of 100s of Gt C over the next few centuries, with similarly significant effects on the atmosphere. Critical areas for future research are continued measurements and analyses of atmospheric data (CO sub(2) and super(13)CO sub(2)) to serve as large-scale constraints, process studies of the scaling from the photosynthetic response to CO sub(2) to whole-ecosystem carbon storage, and rigorous quantification of the effects of changing land use on carbon storage.