The acceptance and use of either surrogate end points (SEPs) or efficient clinical end points are associated with greater and more rapid availability of new medicines as compared with disease situations for which clinical end points are inefficient or no surrogates exist. This review of the history of the development, qualification, and acceptance of key SEPs shows that both successes and failures had three key characteristics: (i) apparent biologic plausibility, (ii) prognostic value for the outcome of the disease, and (iii) an association between changes in the SEP and changes in outcome with therapeutic intervention--the three factors recommended for SEPs in the International Conference on Harmonisation's "Statistical Principles for Clinical Trials." We recommend that only prognostic value be an absolute prerequisite for surrogacy, because therapeutic interventions may not exist a priori, and biological plausibility can be subjective. Ideally, all three of these factors would be traded off against one another in a consistent and transparent risk-management process.