The species composition of microbial communities in natural habitats may be extremely complex and therefore a quantitative analysis of the fraction each species contributes to the consortium has proven to be difficult. During recent years, the identification of bacterial pure cultures based on their infrared spectra has been established. Fourier-transform infrared microspectroscopy now proceeds a step further and allows identification of microorganisms directly plated from community dilutions. Infrared spectra of microcolonies of 70–250 µm in diameter can be recorded without producing a pure culture of the isolate. We have applied this novel technique for quantitative comparative analysis of two undefined, geographically separated food-borne smear cheese microbial consortia of limited complexity. Due to the high degree of automation, up to 200 microcolonies could be identified in 1 day and, in total, 3170 infrared spectra of microcolonies were recorded. The results obtained have been verified by Fourier-transform infrared macrospectroscopy and 16S rDNA sequencing. Interestingly, although the communities were unrelated, Staphylococcus equorum, Corynebacterium casei, Arthrobacter casei and Brevibacterium linens were found to be part of both consortia, however, with different incidence. In addition, Corynebacterium variabile, Microbacterium gubbeenense, Brachybacterium alimentarium, Enterococcus faecalis and an unknown species were detected in either one of the consortia.