Drug selection is now widely viewed as an important and relatively new, yet largely unsolved, bottleneck in the drug discovery and development process. In order to achieve an efficient selection process, high quality, rapid, predictive and correlative ADME models are required in order for them to be confidently used to support critical financial decisions. Systems that can be relied upon to accurately predict performance in humans have not existed, and decisions have been made using tools whose capabilities could not be verified until candidates went to clinical trial, leading to the high failure rates historically observed. However, with the sequencing of the human genome, advances in proteomics, the anticipation of the identification of a vastly greater number of potential targets for drug discovery, and the potential of pharmacogenomics to require individualized evaluation of drug kinetics as well as drug effects, there is an urgent need for rapid and accurately computed pharmacokinetic properties.