In many situations, one would like to be able to rapidly detect the presence of some microorganisms of interest in a given sample. For armed forces in operation for instance, it may be crucial to swiftly detect the occurrence of highly pathogenic germs aerosolized by the enemy. Though exquisitely selective and sensitive, molecular biology based assays are too slow to fulfil this requirement whereas spectroscopic techniques used in the currently available sensors intrinsically suffer from lack of selectivity and/or sensitivity. Numerous reports suggest that it may be feasible to extract robust signatures of bacteria species from their Raman spectra, and that these signatures may be used to identify them at the species level. However, all of theses studies have relied on a rather small number of microorganisms grown in the laboratory under well defined culture conditions, and it is conceivable that the classification models built up with this kind of setting may be overfitted and may behave poorly indeed when confronted with environmental samples with much higher biodiversity. The goal of the SIBIRAM project is to build up a chemometric model based on the Raman signature of microorganisms found in natural aerosols and to assess its robustness by applying it for the description of the microbiological flora of various aerosols at the species level. To that end, we will first acquire the Raman spectra of individual microorganisms immediately after their collection with an air sampler. We will then identify the species of each of them by ribotyping and use these data to build up and optimize the chemometric model using various multivariate analysis methods. We will then use the model to characterise natural aerosols that will have been contaminated with a harmless bacteria species in order to simulate the presence of a pathogen. We will thus be able to assess the performance level of this tool in terms of selectivity and sensitivity and compare it with that of an ideal biosurveillance sensor. In order to increase this performance level, we will design a new protocol with the aim of increasing the species-specific information content of the Raman spectra. To that end, we will extend the spectra in a second dimension by recording them using several excitation wavelengths. We will also stress the microorganisms in several ways prior to spectra acquisition. By sweeping the stress parameters, we will end up with multidimensional spectra that we will use to upgrade our chemometric model. Using this model to describe the microbiological content of aerosols at the species level, we will be able to assess the new level of performance and the added value of our multidimensional Raman scattering approach. Finally, we will use these results to define some key technical requirements that a Raman scattering based biosurveillance sensor should fulfil.