Drug development attrition rates are rising and late phase attrition remains high, contributing to an unsustainable increase in R&D spending. Consequently, there is much effort to identify the potentially successful molecules earlier in development with the use of biomarkers to predict potential efficacy and safety. However, focussing only on picking the winners earlier will not solve the problem. It is essential that the evaluation of these biomarkers also enables the earlier termination of the molecules that will not have the required activity.