Abstract. The purpose of this study was to investigate the potential of near infrared spectroscopy combined with multivariate analysis for determination of bacterial species, caused mastitis infection, based on NIR spectra of milk. Experimental mastitis was induced in 12 milking rabbits. The rabbits were infected with Staphylococcus aureus and E.coli bacteria, isolated from mastitis cows by injection of 0.5 ml of bacterial suspension with different concentration into the base of the teats. After the inoculation of bacterial strains into the teats, the infected rabbit milk was collected at different time intervals – 24, 48, 72, 96, 120 and 144 hours. Spectra of diluted milk were collected using a USB4000 visible-near-infrared spectrometer (OceanOptics, USA) over the wavelength range 450–1100nm using transmission through 10mm quvette. The instrument was first set up with healthy milk as reference. Spectra of 37 milk samples from rabbits, contaminated with Escherichia coli, and 28 milk samples from rabbits, contaminated with Staphylococcus aureus, were used in the investigation. Soft Independent Modeling of Class Analogy (SIMCA) was implemented to create models for discrimination of milk according to bacterial infection. SIMCA models correct classified from 81.08 to 100% of milk samples from rabbits, infected with Escherichia coli, and from 89.28 to 100% of samples from rabbits, infected with Staphylococcus aureus, depending on used spectral region and spectral data transformation. Models, based on spectral region from 456 to 960 nm allowed 100% correct identified all samples. The information of SIMCA models was used for investigation of spectral information, related to presence and action of Escherichia coli and Staphylococcus aureus bacteria in milk. The most important spectral region for detection of Escherichia coli infection was found to be 720 – 750nm, and for Staphylococcus aureus infection – from 920 to 960nm, respectively. The results demonstrated that near infrared spectroscopy in combination with multivariate chemometrics technique offers an alternative approach to traditional methods with large potentials for a rapid and reliable identification in microbiology and biodiagnostics.