n a previous investigation a phenomenological model, based on combined stability/instability indices, was set up to predict stability of extra virgin olive oil (EVOO). According to the model proposed, degradation of EVOO can be monitored by measuring parameters (degradation parameters) which have to be sensitive to oil degradation. From this phenomenological model the following predictive hypothesis was set up to develop a mathematical model for virgin olive oil degradation prediction: the same combination of stability/instability index values corresponded to the same degradation rate, or rather different stability/instability combinations corresponded to different degradation rates. The aim of this study was to confirm the The aim of this study was to confirm the validity of this phenomenological model, to study in depth and to confirm the predictive hypothesis described. The degradation level of a sample series, optimized by a central composite design, was analyzed. Statistical data processing, carried out by partial least squares (PLS), allowed us to confirm the validity of phenomenological model and to improve it. Subsequently, a shelf-life test was carried out. Statistical data processing, carried out by principal component analysis (PCA), showed that degradation parameters significantly explained the degradation rate. In particular, K 270 value and tyrosol and hydroxytyrosol content had a regular, non-asymptotical behaviour, and, as a result, a kinetic study for these parameters could be obtained. These results showed that the predictive hypothesis was likely to be true and suggested that some new shelf-life tests would be required to further investigate our predictive stability model.