In September 2007, a small test-of-concept efficacy trial (STEP) of an HIV vaccine candidate developed by Merck was stopped early for futility. Results from subsequent exploratory analyses of the relatively small amount of data from this trial has had a profound impact on the field of HIV vaccine research and development. Specifically, a companion efficacy trial being conducted in Africa (Phambili) of the Merck vaccine candidate was immediately stopped, initiation of a large efficacy trial testing a different but related vaccine candidate was stopped and a complete re-examination of the scientific strategies and funding of HIV vaccine research was performed by NIH. In this talk, I will provide a brief description of the study design and present results from the primary, secondary and key exploratory analyses of data from the STEP trial. I will then discuss some of the decisions that were motivated by the STEP trial results and describe the relative roles in these decisions of formal statistical inference and of non-statistical judgments about generalizability of results. Questions will be raised about the appropriate role of statisticians in this process and potential implications for statistics training programs.