The introduction of prostate-specific antigen (PSA) as a reliable tumor marker for prostate cancer brought significant changes in endpoints after therapy and in outcome reporting. Over the last 15 years we have collected follow-up information in this new era and struggled with failure definitions using this new tool. Parameters for failure after radiation were especially controversial due to the fact that, unlike surgery, a variable amount of normal prostate function and PSA production remained. In 1996, the ASTRO Consensus Conference established a PSA failure definition based on the available information at the time. It was commonly used for outcome reporting subsequently although criticisms have been voiced and alternate definitions proposed. A recently assembled multi-institutional database was used both for long-term outcome reporting with external beam radiation and to test various other failure definitions. A summary of these results and the associated issues are presented here.