The Syrian hamster embryo (SHE) cell transformation assay (pH 6.7) has utility in the assessment of potential chemical carcinogenicity (both genotoxic and non-genotoxic mechanisms of action). The assay uses morphological transformation as an end point and has a reported sensitivity of 87%, specificity of 83% and overall concordance of 85% with in vivo rodent bioassay data. However, the scoring of morphologically transformed SHE cells is subjective. We treated SHE cells grown on low-E reflective slides with benzo[a]pyrene, 3-methylcholanthrene, anthracene, N-nitroso-N-methylnitroguanidine, ortho-toluidine HCl, 2,4-diaminotoluene or D-mannitol for 7 days before fixation with methanol. Identified colonies were interrogated by acquiring a minimum of five infrared (IR) spectra per colony using attenuated total reflection Fourier-transform IR spectroscopy. Individual IR spectra were acquired over a spatial area of approximately 250 x 250 mu m. Resultant data were analysed using Fisher's linear discriminant analysis and feature histogram algorithms to extract classifying biomarkers of test agent-specific effects or transformation in SHE cells. Clustering of spectral points suggested co-segregation or discrimination of test agent categories based on mechanism of action. Towards transformation, unifying alterations were associated with alterations in the Amide I and Amide II peaks; these were consistently major classifying biomarkers for transformed versus non-transformed SHE cells. Our approach highlights a novel method towards objectively screening and classifying SHE cells, be it to ascertain test agent treatment based on mechanism of action or transformation.