The impetus for this discussion is a Phase III Non-Small Cell Lung Cancer study. (1) The study originally utilized an adaptive design, but it became group sequential following discussion with the FDA. When is it advantageous to propose an adaptive design and would it have been better to begin with the group sequential design which it has since become? (2) The trial was powered based on assumptions including median survival for both treatment and placebo group but it may be that both groups have twice the assumed length of survival. Is it still clinically relevant to have the same assumed difference? (3) What is the effect if the actual treatment effect is twice what was initially thought for both groups?