The application of drug metabolism expertise to early compound selection and optimization has reduced attrition in human pharmacokinetic studies. This reduction has been primarily driven by an increased understanding of the physicochemical properties required in order for a compound to exhibit an appropriate human pharmacokinetic profile. Human pharmacokinetic predictions based on preclinical data are often used to select compounds for further progression. However, the level of prediction accuracy of this approach suggests that the state of the art in human pharmacokinetic prediction will not drive a further reduction in human pharmacokinetic attrition rates. An overall success rate of 60 to 80% of compounds being retrospectively predicted within +/- 2-fold of actual human parameters is insufficient to discriminate closely related analogs within a series. In addition, the post genomic era has led to an explosion in pharmacological targets requiring physicochemistry for interaction with the target that is outside that required for desirable ADME properties. Such targets drive selection decisions into a space where actual human pharmacokinetics are complex and the results from human pharmacokinetic prediction methods are therefore at their most variable. Consequently, the most effective way to operate in these more complicated situations is to devise a rapid and low-cost strategy to complete low-dose human pharmacokinetic studies.