GRIND-based 3D-QSAR methods are widely used in modern medicinal chemistry, since they are alignment-independent and almost completely automated. Nevertheless, their efficacy in predicting different biological activities for a single data set of compounds remains to be explored. In this study we explore the capabilities and limits of ALMOND procedure to predict the inhibitor potency of a series a non-terpenoid squalenehopene cyclase (SHC) inhibitors, and compare the results with recently published results concerning oxidosqualene cyclase (OSC) inhibitor potency. The findings show that the ALMOND procedure can correctly predict both activities, despite the similar architecture of the active center cavities of the two enzymes. Moreover, the graphical results suggest that a compound to act as an OSC inhibitor should satisfy more structural requirements than those necessary to be successful as an SHC inhibitor.