In the era of cytotoxic therapies, tumor regression has rarely been observed in phase I trials and randomized controlled trials have usually been required to demonstrate modest improvements over prevailing standards of care. In the era of effective targeted therapies, drugs such as vemurafenib and crizotinib have demonstrated convincing efficacy in early clinical testing, raising the question of whether randomized phase III trials are necessary and feasible before drug approval. Since 1992, the FDA has approved a number of drugs without data from confirmatory clinical trials as part of the accelerated approval process. While this initiative has largely been successful in bringing effective drugs to the market more quickly, there is much to be learned from case studies of drugs, such as gefitinib, which subsequently failed to gain full approval. In this Review, we use a number of historical examples to make the case that it may be reasonable to consider foregoing randomized phase III trials for certain drugs before drug approval. We explore the consequences (both good and bad) of foregoing randomized phase III trials and propose criteria that might be used to select drugs for consideration of such an approach.