As part of a comprehensive study on tropical land use change and its effect on atmospheric trace gas fluxes, we report the CO fluxes recorded at a natural grassland site and the changes produced when this ecosystem was managed or cultivated. The field site is located in the central part of the savannah climatic region of Venezuela. Fluxes were measured in the dark using the enclosed chamber technique. CO was analyzed with a reduction-gas detector in combination with a molecular sieve 5A column for CO separation. At all sites, CO fluxes exhibited a strong diurnal variation, with net emission during daytime and consumption or no fluxes during nighttime. In unplowed soils no differences were observed between dry and rainy season. A large disparity was observed between unplowed and plowed grassland soils. Plowed soil shows a much smaller emission during daytime and a larger consumption at night. The 24-hour integrated fluxes indicate that the nonperturbed grassland switches from being a net source of CO (3.4¿1010 molecules cm-2 s-1) to being a net sink (-1.6¿1010 molecules cm-2 s-1) after plowing. It is likely that burial of surface litter reduces the production of CO in the top soil and that the diffusion of CO to deeper layers (where CO is consumed by microbiological processes) is promoted in decompacted soils. As the rainy season progressed the plowed soil gradually compacted and CO fluxes changed back, and after 3 months the fluxes from plowed soils and the original unployed soils were equal. Even though the various cultivated fields (corn, sorghum, and pasture) received differing inorganic fertilization treatments, no significant difference in the CO fluxes resulted. Measurements during the dry season suggest that ''degrading dry (dead) vegetation'' produces CO under dark conditions. ¿ American Geophysical Union 1994