AbstractBACKGROUNDCalibration models for the Fourier transform–near infrared (FT‐NIR) instrument were developed for quick and non‐destructive determination of oil and fatty acids in whole achenes of milk thistle. Samples with a range of oil and fatty acid levels were collected and their transmittance spectra were obtained by the FT‐NIR instrument. Based on these spectra and data gained by the means of the reference method – Soxhlet extraction and gas chromatography (GC) – calibration models were created by means of partial least square (PLS) regression analysis.RESULTSPrecision and accuracy of the calibration models was verified via the cross‐validation of validation samples whose spectra were not part of the calibration model and also according to the root mean square error of prediction (RMSEP), root mean square error of calibration (RMSEC), root mean square error of cross‐validation (RMSECV) and the validation coefficient of determination (R2). R2 for whole seeds were 0.96, 0.96, 0.83 and 0.67 and the RMSEP values were 0.76, 1.68, 1.24, 0.54 for oil, linoleic (C18:2), oleic (C18:1) and palmitic (C16:0) acids, respectively.CONCLUSIONThe calibration models are appropriate for the non‐destructive determination of oil and fatty acids levels in whole seeds of milk thistle. © 2014 Society of Chemical Industry