Many approved therapies are commonly used off-label. However, when academic sponsors seek to compare commonly used off-label treatments head-to-head, an IND from FDA is necessary. In such comparative effectiveness research (CER) trials the research question and design strategy often differ from a typical registration study. For example, a placebo is frequently not ethical. Also the definition of a Type I error is less clear in a 3 or more arm trial that is not placebo-controlled. Furthermore a typical research question is "which treatment is best", not necessarily a question of statistical hypothesis test. Finally the primary aims may be multivariable, including simultaneous considerations of effectiveness (not necessarily efficacy), safety, and quality of life. Each of these considerations differs from standard industry-sponsored registration trials. In this session we propose an academic investigator involved in NIH-funded CER trials to discuss goals of such CER trials and an FDA reviewer to discuss how FDA treats (or doesn't treat) these trials differently from industry-sponsored trials seeking labeling.