Carbon sequestration in soils basically depends on both the fluxes of organic carbon from the primary production into the soils and the residence times inside the soil. Up to now, the quantitative dynamic representation of soil carbon, e.g., through simulation models, has been poorly related to the chemical representation of soil carbon. There is a clear need to conciliate both representations for at least two reasons. On one hand, the residence times of carbon in soils are suspected to be intimately related to the chemistry of soil organic matter (SOM), through intrinsic recalcitrance of molecules or organo-mineral interactions. On the other hand, the chemical nature of bulk SOM is intrinsically dynamic and determined by dynamic parameters such as flows of biosyntheses and compound specific reaction rates, including humification and biodegradation etc. Stable isotopes have proven their efficiency tools to investigate dynamics of soil C. At present, compound-specific stable isotope analysis, e. g., using Gas-chromatography/Isotope ratio monotoring mass spectrometry (GC-IRMS) offers the potential to trace the fate of carbon in various classes of molecules. Using these techniques, we built a model of the dynamics of carbohydrate carbon as a part of total soil carbon. We measured the dynamics of carbon in neutral carbohydrates using the natural 13C labelling in an experimental wheat/maize sequence extending from 1 through 23 years. The isotopic composition of individual neutral monosaccharides was determined in hydrolysed particle-size fractions by GC-IRMS of trimethylsilyl derivatives. In each size class, the age distribution of neutral sugar carbon was very similar to that of total soil carbon: few years in particulate organic matter and about one century in <50um particle size fraction. Considering the typical lability of carbohydrates, the relatively high age of carbohydrate carbon may be explained by physical or chemical protection from degradation, as well as by recycling of soil organic matter carbon by soil microbes. The rate of biosynthesis of microbial sugars was determined experimentally using various 13C-labelled substrates and medium term incubation. Based on these experiments we proposed a model of carbohydrate dynamics as a sub-part of a general carbon model. This model respects the frame of a validated model of soil organic carbon dynamics, the RothC model. It introduces specific parameters concerning individual plant carbohydrates degradability, microbial carbohydrates biosyntheses and carbohydrates preservation in soils. Glucose and xylose are more labile than other plant carbohydrates, microbial biosyntheses are independent on the substrate quality, carbohydrate accounting for 30% of the metabolites and the preservation of carbohydrate was about 5 times less important that the one of bulk SOM. The model reproduces nicely the great features of carbohydrate fate in soil i.e. the decrease of their contribution to soil organic matter and the increase of microbial sugars signature during the decomposition process. The proportion of microbially-mediated carbohydrate carbon represents 25% of total soil carbohydrate. Such modeling offers new perspectives in the mechanistic approach of the carbon sequestration in soils. References: Jenkinson, D.S. & Rayner, J.H. 1977. The turnover of soil organic matter in some of the Rothamsted classical experiments. Soil Science, 123, 298-305. Derrien, D., Marol, C., Balabane, M. & Balesdent, J. in press. The turnover of carbohydrates in a cultivated soil estimated by 13C natural abundances. European Journal of Soil Science. Derrien, D. & Balesdent J. submitted. A dynamic simulation of the carbohydrate component of soil organic matter.