Because metabolic problems lead to too many failures during clinical trials, much effort is devoted to in silico models to predict metabolic stability and metabolites. We focused our attention to the prediction of the hydrolytic activity of the carboxylesterase 1 (hCES1) and 2 (hCES2). The obtained results emphasize some crucial properties of the catalytic cavity of these enzyme, confirming that hCES1 prefers substrates with small alcohol groups and large hydrophobic acyl moieties, whereas hCES2 better accommodates substrates with with large alcohol groups and small acyl moieties. Finally, the docking results confirm the relevance of hydrophobic interactions, which seem to govern the hCES1 recognition and allows a robust prediction of hCES1 catalytic activity. Using a substrate trainig set, we found a good relationship between the experimental data and the predicted hydrolysis constants (pKm) .