In clinical trials of oncology drugs, overall survival (OS) is a direct measure of clinical efficacy and is considered the gold standard primary efficacy end point. The purpose of this study was to discuss the difficulties in using OS as a primary efficacy end point in the setting of evolving cancer therapies. We suggest that progression-free survival is an appropriate efficacy end point in many types of cancer, specifically those for which OS is expected to be prolonged and for which subsequent treatments are expected to affect OS.