As a critical parameter in pharmacokinetics, prediction of clearance is an integral aspect of drug discovery programmes. Since the liver is the major site of xenobiotic biotransformation, accurate prediction of hepatic clearance (CLh) is vital. The use of cellular and subcellular in vitro systems for this purpose is common practice; however, prediction accuracy tends to be poor. The aim of this thesis was to explore potential contributing factors to the underprediction of in vivo clearance, specifically with relation to the in vitro methodology of hepatocyte clearance assays, which is largely unstandardised.Literature data analyses highlighted an overall clearance-dependent trend of underprediction in both human and rat hepatocytes, indicating a fundamental in vitro system bias which is independent of species. During initial investigation of incubation conditions, the format of hepatocyte assays (suspension in microcentrifuge tubes, 96-well plates, 24-well plates and short-term monolayer culture) was demonstrated to influence determined intrinsic clearance (CLint). Differences in midazolam CLint were observed not only between suspended and short-term cultured hepatocytes, but also between suspended hepatocytes in different vessels/plate formats. The applicability of 1 µM as a generic substrate incubation concentration for determination of CLint by substrate depletion was evaluated in rat hepatocytes using nine well-characterised drugs. For seven of the nine drugs, a statistically significantly (p < 0.05) higher CLint was observed in determinations of 0.1 µM substrate as opposed to 1 µM, highlighting the potential for false determinations using current practices.Cofactor depletion in isolated hepatocytes was investigated based on previous speculation as the cause of clearance-dependent underprediction of in vivo clearance. Although moderate increases in CLint were observed with the addition of NADPH to hepatocyte incubations, this was subsequently attributed to the replenishment of NADPH in membrane-damaged hepatocytes. Retained functionality of metabolic enzymes in cells which would generally be considered non-viable by trypan blue exclusion was indicated in comparisons of unpurified and Percoll-purified cryopreserved hepatocytes. This phenomenon was conclusively demonstrated in incubations of permeabilised hepatocytes supplemented with NADPH, revealing a need for re-evaluation of the use of plasma membrane integrity (trypan blue exclusion) as a measure of viability in metabolic studies. ATP content was considered as a potential alternative measure; however no significant correlations were found between ATP content, trypan blue exclusion and the CLint of nine drugs in associated preparations.The effect of shaking on CLint in rat hepatocytes was also examined. For 10 out of 12 drugs, CLint determined at 900 rpm was significantly (p < 0.05) higher than in static incubations. Three potential mechanisms were hypothesised: plasma membrane damage, increased substrate distribution throughout the bulk medium and reduction in the depth of the unstirred water layer (UWL) surrounding cells. Shaking of saponin-permeabilised hepatocytes (supplemented with NADPH to maintain metabolism) also increased the determined CLint of saquinavir, indicating a rate-limitation other than membrane permeation. However, shaking of ultra-sonicated hepatocytes in which the plasma membrane was entirely destroyed (also supplemented with NADPH) did not change the determined CLint of saquinavir, revealing the rate-limitation of UWL permeation in both intact and permeabilised cells. The depth of such an UWL in vitro is likely to be artificially greater than in vivo; therefore reduction of UWL depth through incubation shaking is proposed as a physiologically sound approach by which to increase in vitro CLint. In addition, a framework of experiments and related equations is presented by which intact and permeabilised hepatocytes in static and shaken conditions may be utilised to identify the rate-determining process and contribution of individual processes to the in vitro CLint of a drug. The effects of substrate concentration and shaking were also evaluated in human hepatocytes. Significant increases in determined CLint of drugs with use of 0.1 µM substrate (as opposed to 1 µM) and shaking at 900 rpm were demonstrated, confirming equivalent potential in vitro sources of underprediction between rat and human.These highly significant findings reveal the existing limitations of in vitro assays and potential flaws in current practice in in vitro determinations of CLint. Appropriate consideration of the properties of in vitro systems, including the presence of the UWL, should lead to improved predictions of in vivo clearance.