Natural grasslands, especially in the tropics, urgently need more detailed study in order to determine the response of this undervalued major ecosystem type to possible climate changes. Feedback effects through environmental variables such as temperature, water and nutrient stress may be at least as significant as the increase in atmospheric CO2 concentration, but there is scarcely enough data at present to develop and validate modelling. Annual burning of large areas of tropical grasslands plays a significant role in the global carbon cycle. Net loss of soil carbon and nitrogen may result, depending upon the frequency of fire, overgrazing and drought. The UNEP Project on productivity and photosynthesis in tropical grasslands attempts to correct the gap in baseline data, and has found these ecosystems to be far more productive than previously appreciated. Based on data from three terrestrial grassland sites, the gross flux of carbon from burning of tropical grasslands falls in the range 2.4-4.2 Gt per annum, a significant amount compared with the net fluxes of 1.8 Gt estimated from deforestation and 5.3 Gt from fossil combustion. Data from this project is also being applied to modelling work in collaboration with SCOPE, in order to study climate change effects on carbon cycling in grasslands. © 1991.