The diagnosis and management of inflammatory bowel disease relies on histological assessment, which is costly, subjective, and lacks utility for point-of-care diagnosis. Fourier-transform infra-red spectroscopy provides rapid, non-destructive, reproducible, and automatable label-free biochemical imaging of tissue for diagnostic purposes. This study characterises colitis using spectroscopy, discriminates colitis from healthy tissue, and classifies inflammation severity. Hyperspectral images were obtained from fixed intestinal sections of a murine colitis model treated with cell therapy to improve inflammation. Multivariate analyses and classification modelling were performed using supervised and unsupervised machine-learning algorithms. Quantitative analysis of severe colitis showed increased protein, collagen, and nucleic acids, but reduced glycogen when compared with normal tissue. A partial least squares discriminant analysis model, including spectra from all intestinal layers, classified normal colon and severe colitis with a sensitivity of 91.4% and a specificity of 93.3%. Colitis severity was classified by a stacked ensemble model yielding an average area under the receiver operating characteristic curve of 0.95, 0.88, 0.79, and 0.85 for controls, mild, moderate, and severe colitis, respectively. Infra-red spectroscopy can detect unique biochemical features of intestinal inflammation and accurately classify normal and inflamed tissue and quantify the severity of inflammation. This is a promising alternative to histological assessment.