A new identification method for bifidobacteria species from the human gastrointestinal tract was developed based on the measurement and statistical analysis of the intrinsic fluorescence of aromatic amino acids (AAA) and nucleic acids (NA), following their excitation at 250 nm. The model was constructed by recording the fluorescence spectra of 53 Bifidobacterium strains of 10 different species, including the corresponding type strains, and validated by analyzing the spectra data from nine further problem strains. Principal components analysis (PCA) and factorial discriminant analysis (FDA) of the results showed the technique to distinguish between the isolates at the species level; the Bifidobacterium pseudolongum subspecies (globosum and pseudolongum) could also be distinguished. The proposed method provides a powerful, inexpensive and convenient means of rapidly identifying intestinal bifidobacteria, which could be of help for large probiotic surveys.