Modelling of soil respiration plays an important role in the prediction of climate change. Soil respiration is usually divided in a fraction originating from root respiration and a heterotrophic fraction originating from microbial decomposition of soil organic carbon. This paper reports on the coupling of an one-dimensional water, heat and CO2 flux model (SOILCO2) with a pool concept of carbon turnover (RothC) for the prediction of soil heterotrophic respiration. In order to test this coupled model, it was applied to a bare soil experimental plot located in Bornim, Germany. Soil temperature and soil water content measurements were used for comparison with the respective model predictions. An 8 years data set Of CO2 efflux measurements, covering a broad range of atmospheric conditions, was used to evaluate the model. In a first step we quantified the improvement of the CO2 efflux prediction due to the coupling of the flux model with a pool concept of carbon turnover. The humus pool decomposition rate constant and its soil water content dependent reduction were derived from the first 5 years Of CO2 efflux measurements using inverse modelling. The following 3 years of measurements were used to validate the model. The overall model performance Of CO2 efflux predictions was acceptable with the measured and simulated mean daily respiration being 0.861 and 0.868 g C m(-2) d(-1), respectively, and a mean absolute difference between modelled and measured rates of 0.21 g C m(-2) d(-1). The inverse estimation of the humus decomposition rate constant resulted in a value of 0.04 year(-1), which is higher than the default value in RothC. This is attributed to the agricultural practice during the experiment. (c) 2008 Elsevier B.V. All rights reserved.