This study was performed to explore the feasibility of using the visible spectral region (360-740 nm) for the rapid and non-destructive prediction of lipid oxidation in rabbit meat stored at 4 +/- 1 degrees C for 1, 6 and 12 days. The analyses were performed on vacuum packed meat stored in darkness at 4 +/- 1 degrees C on day 1, 6 and 12. Spectra were analysed through multivariate regression (PLSR) calibration model, using the full spectral region between the pre-treated spectra by second derivatives and the reference TBARS values. The high coefficient of determination (R-2 = 0.87) between the predicted and measured TBARS values as well as the high values of the root-mean-square errors of calibration and prediction (RMSEC = 0.478 and RMSEP = 0.532) suggested the usefulness of visible spectroscopy in predicting meat oxidation. Furthermore, the suitability of five wavelengths (380, 420, 440, 580 and 600 nm), selected by PLSR, to discriminate meat samples according to storage times, was also investigated by discriminant analysis performed on a new data set, composed of the second derivative spectral data, measured on meat samples (n = 900), which were collected from local supermarkets at 3 different storage times (1, 3 and 6 days). Overall accuracy was 82%, thus corroborating the feasibility of using visible spectroscopy to assess the qualitative decay in rabbit meat. (C) 2015 Elsevier Ltd. All rights reserved.