Comparison studies have the potential to do as much harm as good in the field of applied behavior analysis. The following review identified 62 studies in the past five years that compared which of two treatment parameters was better for individuals with autism (e.g., massed trials vs. interspersed trials; picture exchange vs. manual signs; etc.). In nearly half of those studies, all participants progressed further with one treatment procedure. In the other half of the studies, results were mixed. Some clients progressed further with one treatment procedure while other clients progressed further with the other treatment procedure. While the former studies are interesting, this paper argues that the latter studies (with mixed results) do more to advance the practice of behavior analysis. In addition, this paper explores a) which procedural components have most commonly been compared, b) potentially relevant client characteristics associated with different results, and c) discrepancies between treatment in the research and treatment in practice. Dr. Lovaas once advised, "If a child cannot learn in the way we teach, we must teach in a way the child can learn." Rather than focusing on which treatment is superior, this paper encourages more nuanced practice decisions from behavior analysts.