||Soil CO2 flux (J(CO2)) was measured at midday over a 2-yr period in undisturbed tallgrass prairie (Konza Prairie, Kansas, USA) to quantify seasonal and annual budgets, to evaluate temperature and moisture as determinants of soil CO2 flux, and to assess the effect of a common land management tool, spring fire, and fire history on soil respiration. We hypothesized that: (1) maximum rates and annual estimates of soil J(CO2) would be greater in more productive burned sites than in unburned sites, (2) soil J(CO2) would be greater in newly burned sites with a history of fire exclusion than in annually burned sites (consistent with differences in aboveground production), and (3) soil temperature and water availability would be primary abiotic determinants of soil J(CO2) in tallgrass prairie. A preliminary assessment of the effects of large herbivores on soil J(CO2) was included to evaluate the hypothesis that removal of aboveground biomass would reduce soil J(CO2). Results indicated that spring fire increased maximum monthly soil J(CO2) by 20-55% relative to unburned tallgrass prairie, with greatest monthly differences measured in April (fourfold higher in burned sites). In burned sites that differed in fire history, maximum monthly J(CO2) in annually burned prairie was 33% greater than in burned sites with a history of fire exclusion. Soil J(CO2) in these latter sites was still significantly higher than in unburned sites. Soil J(CO2) in sites grazed by bison was reduced by as much as 30% relative to adjacent ungrazed areas. Reduced root biomass and activity in grazed areas, unburned sites, and sites with a history of fire exclusion suggest that plants play a major role in determining soil J(CO2) in this grassland. Soil temperature at 5 cm was related strongly to midday J(CO2) in both annually burned sites (r(2) = 0.58) and unburned sites (r(2) = 0.71). In contrast, differences in soil moisture among sites, enhanced by comparing irrigated grassland to control areas, increased maximum monthly J(CO2) by only 8%. Thus, soil temperature was the primary abiotic determinant of soil J(CO2) during this study. Maximum monthly estimates of soil J(CO2) in tallgrass prairie ranged from 10.3 mu mol CO2 . m(-2) . s(-1) in unburned sites to 15.1 mu mol . m(-2) . s(-1) in annually burned irrigated sites, whereas annual estimates varied from 4.7 to 7.8 kg CO2/m(2). Over the 2-yr period, spring fire increased estimated annual soil J(CO2) by 38-51% relative to unburned sites, while irrigation increased annual soil J(CO2) by 13%. These estimates for tallgrass prairie are much higher than those reported for most temperate ecosystems but are similar to estimates for tropical forests. Characteristics of undisturbed tallgrass prairie that may lead to high levels of soil J(CO2) include: high above- and belowground productivity; a relatively high proportion of C stored belowground; levels of soil microbial biomass and activity that are among the highest in native ecosystems in the United States; and the lack of a single dominant factor such as temperature, moisture, or nutrient availability, that consistently limits biotic processes during the growing season. The sensitivity of soil J(CO2) in tallgrass prairie to different land use practices (fire and grazing) suggests that it is critical to include these factors in the development of grassland C budgets, as well as in regional models that estimate biogeochemical responses to land use change.