Multivariate analysis techniques, principal component analysis (PCA), principal component regression (PCR) and partial least square regression (PLSR), were employed to develop calibration and prediction models for the determination of oil yield from oil shale samples using diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS). Data pre-processing included the use of second-derivative spectral data. Multi-component models were constructed and were effective in predicting oil yield with accurate predictions achieved using oil shale samples other than those used in the calibration set. DRIFTS with multivariate calibration modelling is demonstrated to provide a simple and rapid method of evaluating oil yield from oil shales compared with, and potentially replacing, the traditional modified Fisher assay (MFA) method.