The rapid identification of the etiological agent of microbial infections can bring about both clinical and financial benefits. Thus, fast and generally applicable classification methods are needed that will enable us to rapidly distinguish pathogenic bacteria from commensals or saprophytic bacteria found in the same habitat. We here show that provisional classification of bacterial isolates can be performed on a large scale based on 16S rRNA sequence comparisons using Pyrosequencing, a recently described real-time DNA sequence analysis technique, and the concept of signature matching. The probes we have developed, together with the new technology, will enable early diagnosis of specific pathogens, which is critical for the rational use of antimicrobial therapy in clinical medicine.