Abstract. In this work, the rich dataset acquired at the SMOSREX experimental site is used to enhance the A-gs version of the Interactions between Soil, Biosphere and Atmosphere (ISBA) model. A simple representation of the soil moisture effect on the ecosystem respiration is implemented in the ISBA-A-gs model. It results in an improvement of the modelled CO₂ flux over a grassland in southwestern France. The former temperature-only dependent respiration formulation used in ISBA-A-gs is not able to model the limitation of the respiration under dry conditions. In addition to soil moisture and soil temperature, the only parameter required in this formulation is the ecosystem respiration parameter Re₂₅. It can be estimated by means of eddy covariance measurements of turbulent nighttime CO₂ flux (i.e. ecosystem respiration). The resulting correlation between observed and modelled net ecosystem exchange is r²=0.63 with a bias of −2.18 μmol m⁻² s⁻¹. It is shown that when CO₂ observations are not available, it is possible to use a more complex model, able to represent the heterotrophic respiration and all the components of the autotrophic respiration, to estimate Re₂₅ with similar results. The modelled ecosystem respiration estimates are provided by the Carbon Cycle (CC) version of ISBA (ISBA-CC). ISBA-CC is a version of ISBA able to simulate all the respiration components, whereas ISBA-A-gs uses a single equation for ecosystem respiration. ISBA-A-gs is easier to handle and more convenient than ISBA-CC for the practical use in atmospheric or hydrological models. Surface water and energy flux observations, as well as Gross Primary Production (GPP) estimates, are compared with model outputs. The dependence of GPP to air temperature is investigated. The observed GPP is less sensitive to temperature than the modelled GPP. Finally, the simulations of the ISBA-A-gs model are analysed over a seven year period (2001–2007). Modelled soil moisture and Leaf Area Index (LAI) are confronted with the observed surface and root-zone soil moisture content (m³ m⁻³), and with LAI estimates derived from surface reflectance measurements.