Correlation analysis was carried out between binding affinity data values from the literature and physicochemical molecular descriptors of two series of single point mutated canonical inhibitors of serine proteases, namely bovine pancreatic trypsin inhibitor (BPTI) and turkey ovomucoid third domain (OMTKY3), toward seven enzymes. Simple quantitative structure-activity relationship (QSAR) models based on either single or double linear regressions (SLR or DLR) were obtained, which highlight the role of hydrophobic and bulk/polarizability features of mutated amino acids of the inhibitors in modulating both affinity and specificity. The utility of the QSAR paradigm applied to the analysis of mutagenesis data was underlined, resulting in a simple tool to quantitatively help deciphering structure-function/activity relationships (SFAR) of different protein systems.