Several studies have suggested that soil microbial communities adapt to changes in temperature with possible long term consequences for organic matter decomposition. We hypothesised that in response to changes in temperature, microbial decomposition would result in a differential production of metabolites. In order to test this hypothesis we used an approach, called metabolomics, which consists of a non-biased analysis of the metabolites produced by or existing in a cell. It gives a picture of the functionality of a cell or of a microbial population in a given environment. The aim of this study was to demonstrate the reproducibility and the discriminatory power of the analytical approach and then to determine the effect of changes in temperature on metabolite production. Two bacterial strains, Anthrobacter sp. and an Erwinia sp., isolated from soil were incubated in minimal media supplemented with fructose at 20°C. Both the intracellular and extracellular metabolites were extracted and detected by gas chromatography. Analytical variability was compared with biological variability by preparing triplicate liquid microcosms for each strain and then for each step of the protocol. Profiles were compared on the basis of relative abundance of each compounds using Principal Component Analysis. Analysis of variance was used to determine the variance components for each experimental factor (Strain \ Microcosm \ Extraction \ Derivatization). The 2 strains were clearly separated along the first ordination axis. Species replicates showed some variability along the second axis. ANOVA indicated that the biological variability was more important than the analytical variability. Subsequently, incubations were carried out at different temperatures and the metabolic profiles compared. Differences in metabolic profiles were also compared with catabolic profiles obtained with Biolog.