Background: The US Food and Drug Administration (FDA) Accelerated Approval Program (AAP) was instituted to allow patients earlier access to medications that address large unmet need and are approved based on surrogate endpoints. Recently, approvals have been criticized for not providing overall survival data, and for confirmatory trials conducted in alternative populations. We investigated the availability of overall survival (OS) data for cancer accelerated approval (AA) indications and the correspondence between the population providing OS information and that of the original indication to assess the merit of these critiques. Methods: Approval letters and labels of cancer therapy products approved between January 1, 2006 and June 31, 2021 based on their verification status as of June 10, 2022 were identified from Drugs@FDA. Information was extracted irrespective of the final decision. Information about the exact indications and the trials that were used for AA and were cited for RA (confirmatory trial). OS data was searched for both AA and RA trials irrespective of approval dates, linked by NCT numbers identified in labels and letters. When OS was available only from the trial indicated for the RA, level of correspondence between the source of the OS data and the AA indication were described and then categorized in terms of the population characteristics, treatment lines, disease stage and changes in treatment regimen, to indicate level of correspondence. Results: OS data is available for 83% of closed AAs: 50 among the 57 converted and 12 among 18 withdrawn AAs. Further OS data was identified for seven AAs among the 55 ongoing reviews, creating a set of OS data for 69 out of the 130 total number of AAs identified. Exact or very good population overlap is evident in 50% of cases, defined as the availability of the trial reporting results for a population exact or very close to the original AA population; line change down was defined if the RA trial was conducted in an earlier line (without overlap with the original AA trial); and some additional categories were identified. Conclusions: Overall, OS information was available, and in half of the cases it was reported for a population very close to the original AA indication. Likely for ethical and practical reasons, RA trials are conducted in an indication that differs from the original AA indication and changes in the population was with the consent of the FDA. Policy discussions about AAP are incomplete without assessing its impact on survival and real-world considerations for conducting confirmatory trials.