Composting is an aerobic biological treatment used to valorise solid organic residues and to recycle them as a soil improver or as an organic fertilizer. A well-managed composting process can contribute to limit greenhouse gas emissions compared to landfilling and to increase the potential of carbon storage in the soils. Nevertheless, an environmental limit of the treatment may be nitrogen emissions. As composting relies on numerous physical and biological linked phenomena, optimisation of its management could be helped by a numerical model. However existing models have been mainly focused on the description of the fate of the carbonaceous part of organic matter in order to simulate oxygen consumption, temperature rise and organic matter stabilisation along the treatment. On the other hand, the fate of nitrogen along composting has been experimentally studied but less considered in terms of modelling. As the knowledge of the behaviour of nitrogenous molecules is necessary 1/ to simulate nitrogen emissions and final quality of the compost and 2/ to use the results as entry data for modelling the behaviour of compost organic matter in soil, the aim of this study was to propose modelling assumptions for nitrogen behaviour along aerobic degradation, considering a fractionation of organic matter based on the Van Soest method. The aerobic biodegradation of four types of waste was studied: fattening steers solid manure (MAN), the solid fraction of digested fattening steers solid manure (DMAN), a mixture of biowaste and green waste (BIOD), the solid fraction of the digested mixture of biowaste and green waste (DBIOD). The oxygen consumption kinetics (OUR) was measured through a respirometric method. For each substrate, several respirometric cells were run at the same time but stopped at different dates along the biodegradation. When stopped the content of each cell was sampled and submitted to a hot water extraction. The nitrogen and carbon content of the water extract (SOLW) was analysed. The water-extracted solid residue (NSOL) was then successively extracted according to the Van Soest fractionation method. N and C content was analysed on each extraction residue so that N and C content in the four fractions SOLNDF (organic matter soluble in neutral detergent), HEM (hemicellulose–like), CEL (Cellulose-like) and LIG (lignin-like) could be calculated. Results showed that the quantity of nitrogen lost along the aerobic biodegradation ranged between 3 and 64 % of the initial N content and varied depending on the origin of the substrate. N was mainly lost from the SOLW fraction, containing ammonia-N and from the SOLNDF fraction, whereas C losses were mainly attributed to HEM and CEL fractions. In terms of kinetics, it appears that up to the peak of OUR, N losses were mainly attributed to the SOLW fraction. After this peak, N losses concerned mainly the SOLNDF fraction which N content decreased and seemed to supply the N content of the SOLW fraction. By the end of the biodegradation a net disappearance of N content in the SOLW fraction was observed again. Moreover, for a substrate as BIOD which N content was mainly contained in NSOL, a transfer of N from HEM and CEL fractions to SOLNDF fraction was observed concomitantly to C hydrolysis of these fractions. The experimental observations allowed us to propose N modelling assumptions. Three solid fractions of organic N are considered: N-HEMCEL, N-LIG, and N-SOLNDF. Two N-water-soluble fractions are considered: organic N-SOLW and mineral-N-SOLW. Hydrolysis of HEM and CEL fractions supplies organic N to fraction SOLNDF which hydrolysis releases organic N molecules to fraction SOLW. This organic N- SOLW is then mineralised to mineral-N-SOLW and used by microorganisms for their metabolism, transformed to N2 (and eventually N2O), or volatilised as NH3. The originality of this work is to demonstrate that, considering a same fractionation method, C and N behaviours in each fraction are kinetically independent which implies to not consider constant C/N ratio in the model. The numerical model has still to be validated on experimental results.