The development of novel drugs falls into two completely different categories: truly novel drugs and drugs that can be considered as improvements of further advanced and eventually marketed drugs. The risk of failure and the reason for failure by these two classes of compounds obviously are very different. Truly novel drugs often rely on pharmacological data obtained in preclinical models paired with a scientific rationale for a mechanism thought to be relevant for the phenotype of the disease. The scientific insight into both the mechanisms underlying the disease and how these diseases can be manipulated by pharmacological means is therefore essential for the success of the drug. In practical terms, this means that a thorough understanding of the disease is a prerequisite for success. It is therefore a sobering thought that most of these compounds fail due to marginal efficacy in Phase II or III trials. The lack of success of these compounds may reflect either lack of knowledge of the disease, poor predictive value of the preclinical models, large heterogeneity in the underlying mechanisms for a given phenotype, or the use of the compound in a population that doesn't express a phenotype optimal for the drug. This year's list of discontinued compounds spans the range from truly innovative drugs to 'me-too' compounds and as such is highly useful in illustrating the current dilemmas for the pharmaceutical industry.