Organic matter quality i.e. the intrinsic biochemical properties of crop residues, strongly influence their decomposition in soil and the associated C and N fluxes. The C-to-N ratio was well established by the 1920s as a potential indicator of residue degradability. Later, in the 1960s, proximate analyses (Van Soest extraction) developed to estimate forage digestibility in the rumen have been adapted to characterize residue quality in addition to the C-to-N ratio. These analyses were used to predict residue biodegradation in soil assuming that there was a parallel between an increasing strength of chemical extraction and a decreasing rate of decomposition. However most of the results published on the relationships between residue quality, define as Van Soest fraction's or C-to-N ratio, and residue decomposition, have underlined the difficulty to predict C and N fluxes. If it is relatively well established that, on the first days following the addition of crop residues to soil, carbon (C) mineralization is mainly correlated to the amount of C present in the soluble fraction of residues (Hermann et al., 1977), explanation and prediction of C mineralization kinetics on longer period of time remained hazardous (Jensen et al., 2005). For instance, root carbon mineralization, which constitutes one of the major source of C to soil organic matter, is not well understood and hardly simulated with C and N models (Abiven et al., 2005). It is likely that the difficulties encountered to predict long term evolution of C in soils, not only result from a lack of data to well estimate the amounts of C involved in the decomposition process but also from the lack of understanding of the importance of residue biochemistry on the decomposition process itself (Bertrand et al., 2005). In this context, our work aimed at 1) better understand the role of residue composition i.e. cell wall network on the decomposition process in soils 2) estimate the relative contribution of cell wall composition and architecture on the decomposition process 3) find new quality criteria to describe residue quality in C and N models. To answer some of these questions two plants were used as model: wheat (c.v. Shango) and maize. In addition to C-to-N ratio and Van Soest extraction, cell wall composition was determined on selected plant parts. The nature and amount of cell wall neutral sugars, of lignin (condensed and non condensed forms) and monomers and of phenolic acids were determined before and after their decomposition in soil. Microscopic observations were performed to increase our understanding of the role of tissue architecture on the decomposition process. Plant parts (roots, leaves, internodes) were incubated in controlled conditions of temperature and moisture. The soil and residues were placed in plasma flask with a CO2 trap and mineral nitrogen was added to ensure that N would not limit the decomposition (Recous et al., 1995). The kinetics of C and N mineralization were determined and, at the end of the incubation, residues were separated from the soil and biochemical analyses were performed on the decomposed residues when possible. As expected, wheat leaves, which were poorly lignified, decomposed faster in soil than internodes and roots (Figure 1). Leaves, also presented a highly condensed lignin structure and the extent to which uncondensed leaf lignin was affected by soil decomposition suggest that the contribution of leaf lignin to C mineralization during incubation was very low. Roots which contained similar amounts of lignin than internodes decomposed more slowly. Roots were enriched in phenolic acids and presented a more condensed lignin structure than internodes (Figure 2). P-coumaric and ferulic acids, the two main forms of phenolic acids in graminceous, have cross link function within the cell wall network that could be of major interest in estimating soil residue degradability. These acids which represent less than 1% of plant dry matter constitute lignin carbohydrates complexes recalcitrant to enzymatic attack. In addition the knowledge of quality of lignin (S/G ratio and level of condensation) could contribute to explain the differences in C mineralization observed for residues containing similar amounts of lignin but decomposing at different rates in soil (roots versus aerial part). The possible role of these cell wall compounds as well as the importance of tissue organization on the decomposition process will be discussed in detail. References: Abiven S et al., 2005 Soil Biol. Fert. 42, 119-128. Bertrand I et al., 2005 Plant Soil, In Press Hermann WA et al., 1977 Can. J. Soil Sci., 57, 205-215. Jensen LS et al., 2005 Plant and Soil, 272, 307-326. Recous S et al., 1995 Soil Biol. Bioch. 27, 1529-1538.