Abstract: In order to estimate the fatty acid composition of low and high melting oils which were fractionated from virgin olive oil, the infrared spectra of 31 oil samples with different fatty acid composition were determined by FTIR spectroscopy in the region 4000-600 cm-1using single reflectance technique. Based on the analysis results, the partial least squares (PLS) calibration model was established. Cross-validated partial least squares calibration models were established between spectral variables and fatty acid composition variables. For the calibration content of oleic acid and linoleic acid, the whole spectral profile in the region 4000–600 cm-1 by smoothing, second derivative, and normalization treatment gave the best cross-validated calibration model and the best prediction results. The predicted results of the PLS calibration models showed that the low melting olive oil had higher contents of oleic acid and linoleic acid (oleic, 77.46%; linoleic, 12.51%) than those of high melting olive oil (oleic, 72.29%; linoleic, 9.98%), which were better consistent with the results of gas chromatography (GC). The established PLS model can predict the main fatty acid composition in fractionated olive oils. This study can provide a simple and rapid analysis method for quality evaluation of fractionated oils.