Rising expenditure in pharmaceutical R&D has not been matched by increased productivity. There is an urgent need to solve the current high levels of pipeline attrition. Changing the current failed model of drug discovery and development, in which high numbers of candidate drugs are produced and high attrition is accepted, is essential. A different model is needed, in which the focus shifts to identifying better-quality candidate drugs that allow scientifically robust testing of disease and targets in humans. Lowering the risks of compound-based attrition in small-molecule drug discovery and development (ie, addressing toxicity, specificity, potency, duration and exposure) is achievable by improved control of physical properties and by setting more demanding candidate criteria. Separating the key scientific experiment--proof-of-concept clinical trials in humans--from commercial development imperatives is a necessary step for the industry.