We investigate the CO2 flux calculated by the ISBA soil-vegetation-atmosphere transfer model (Noilhan and Planton, 1989)by comparing three different formulations for the plant (dark) respiration scheme applied to a soybean culture. The model includes CO2 flux/photosynthesis based on Jacobs (1994) in a manner similar to Calvet et al. (1998) (ISBA-A-gs). The first respiration scheme (M0) computed the autotrophic respiration Rd; similarly to Jacobs (1994) but with an ad-hoc temperature correction calibrated by statistical parameter fitting using measured data. For the second model (M1), we implemented the respiration proposed by Joetzjer et al. (2015). Finally we implemented a third respiration scheme (M2) as in Wang (1996). The three models were calibrated and CO2 fluxes were compared with measurements made over a soybean culture using eddy covariance method between December, 2008 and March, 2009, at a farm near Buenos Aires, Argentina. The total CO2 maximum, minimum and mean measured flux values were respectively 0.9890, -0.2479 and 0.3087 mg m-2 s-1. For the sake of comparison, statistics were computed for the full daily cycle flux (total) and also for nighttime flux, as a means to avoid masking of the results due to the much larger daytime photosynthetic flux. We here present the Nash-Sutcliffe efficiency (NSE) coefficient for each model. M0 gave the best overall performance with 0.7568 for the total daily CO2 flux and 0.0795 for the dark flux. M1 gave similar predictions for the daily CO2 flux with 0.7582, butthe worst result for the nighttime period with -0.4965. M2 gave 0.7424 for the full daily flux and 0.0119 for the night CO2 flux. The results show a seemingly better performance of the models in predicting the total CO2 flux compared to the dark CO2 flux. This is due to several facts such as: respiration is less understood and harder to predict than photosynthesis; measurements are more difficult at nighttime due to the limitations of the eddy-covariance technique in low turbulent activity; in the measured data, it is difficult to identify and separate the portions of CO2 fluxes as soil respiration, autotrophic respiration and photosynthetic flux, without many auxiliary measurements. We also conclude that there is a clear influence of the temperature on the respiration, which can be suitably incorporated in the models.