MetaSite, a software program developed by Molecular Discovery Ltd., that predicts CYP-mediated sites of metabolism based on the CYP enzymeís active site, and orientation of a substrate via GRID molecular interaction field descriptors. It further accounts for accessibility and reactivity of a given CYP and a given substrate respectively, to determine each atomís probability of metabolism1. MetaSite is routinely employed in drug discovery to understand metabolic liabilities of new chemical entities2. The objective of the present investigation was (i) to compare in vitro human liver microsomal metabolite identification results with in silico soft-spot predictions by Metasite and (ii) to understand the molecular and ADME properties leading to high predictive accuracy as well as those drivers implicated in low predictive accuracy. To this effect, human liver microsomal biotransformation studies were carried out for 70 new chemical entities. Top three metabolic soft spots were computed by Metasite, for liver as a metabolic model, with reactivity correction enabled. Comparison between in silico and in vitro results revealed approximately 70% accuracy of prediction by Metasite for new chemical entities. A weighted average of MetaSiteís top three sites of metabolism was calculated from the Metasite score and was compared with human microsomal Clint. Preliminary observations indicate a tendency that the prediction of MetaSite is better for new chemical entities with high microsomal extraction ratios i.e. higher Clint (>300 uL/min/mg of microsomal protein). Similarly a general tendency toward increase in confidence of prediction by Metasite for the top three metabolic soft spots was seen with increase in the number of observed metabolites in human liver microsomal incubations. 1. Cruciani, G.; Carosati, E.; De Boeck, B. et al. (2005) J. Med. Chem. 48, 6970-6979. 2. Trunzer, M.; Faller, B.; and Zimmerlin, A. (2009) J. Med. Chem. 52:329-335.