Carbon dioxide (CO2) release from manure was investigated under held conditions in one of the compartments in a mechanically ventilated fattening pig house. The potential of CO2 release was tested in the emptied compartment by imposing high ventilation rate and measuring CO2 concentrations in the incoming and exhaust air. When the compartment was filled with pigs, a model of tranquil CO2 exhalation rate (TCER) was used to estimate the part of CO2 produced by animals, in addition to the measurement of CO2 concentrations and ventilation rate. Useful data were extracted from 135 days of continuously measured field data obtained when the housing system was in steady-state. The CO2 release per m(2) of pit surface was 18.4 and 25.8 g h(-1) m(-2) in the emptied pig compartment during the impulse and pulse ventilation rate tests, respectively. It ranged from 1.1 and 116.7 g h(-1) m(-2) and averaged 42.1 g h(-1) m(-2) during the 135-day continuous held measurement. In 128 of the 135 days when the compartment was filled with pigs, the mean CO2 release from the manure reached an average of 37.5% of the TCER by pigs: instead of less than 5% believed by many researchers. The maximum CO2 release rate in 2-h steady-state measurements was 94% of the TCER by pigs. This research demonstrated that, in the previous knowledge, there might be a serious underestimation of the CO2 release from the manure in pig houses. The total pig weight, manure temperature and ventilation rate were found the most important variables related to the CO2 release. A mathematical model was developed to estimate the CO2 release as a function of these three variables with R-2 = 0.525. (C) 1999 Elsevier Science Ltd. All rights reserved.