In recent years, compounds eliminated wholly or in part by glucuronidation have represented a significant proportion of Merck's pipeline. However, these compounds generally have uncertainty around their human PK predictions making it difficult to assess the probability of success early in development. Commonly employed in vitro tools to assess metabolism, such as liver microsomes, generally under-predict clearance for compounds eliminated by glucuronidation, more so than for compounds eliminated by oxidative metabolism. In recent studies it was suggested that long-chain unsaturated fatty acids, in particular arachidonic acid, present in human liver microsomes (HLM) can be potent inhibitors of glucuronidation through UGT1A9 and UGT2B7 (Tsoutsikos et al, 2004, Rowland et al., 2007, 2008). In this study we have explored this theory further by quantifying the inhibitory potential of arachidonic acid and other free fatty acids on recombinant human UGT1A9 and UGT2B7. It was found that arachidonic acid is an inhibitor of UGT1A9 and UGT2B7 with IC50,app values of 0.7-0.8 µM and 6-24 µM respectively. Additionally, we have investigated the use of a potent, irreversible Phospholipase A2 (PLA2) inhibitor (methyl arachidonyl fluorophosphonate, MAFP) and bovine serum albumin (BSA) during microsomal preparation to minimize the production of arachidonic acid. HLM prepared from a pool of 3 donors with BSA and MAFP contained almost 7-fold less arachidonic acid compared to control preparations and almost 9-fold less arachidonic acid following incubation at 37°C for 3 h. This translated to an increase in the rate of zidovudine (a UGT2B7 specific substrate) and propofol (a UGT1A9 specific substrate) glucuronidation by 10 and 8-fold respectively in these HLM preparations with no significant difference in microsomal binding between control preparations and preparations containing both BSA and MAFP for either substrate. In vitro-in vivo extrapolations using the well stirred model and data from control microsomes under-predicted human in vivo CLint by more than 100-fold. This under-prediction was improved to 9-fold when microsomes prepared with BSA and MAFP were used. This work provides an alternative microsomal preparation technique that does not rely on altering the commonly employed incubation conditions. As the under-prediction of glucuronidation observed when extrapolating from HLM systems is most likely a multi-faceted issue it is unlikely to be rectified by one explanation alone. However, this work might be one piece of the puzzle towards developing more predictive in vitro tools for predicting human PK of certain compounds eliminated by glucuronidation.