This chapter provides an overview of phase II and III clinical trial designs that utilize predictive markers. After identifying fundamental elements of statistical evaluation of predictive markers, we outline top-down and bottom-up methods as basic adaptive design approaches to evaluate the profile of treatment effects across marker-defined subpopulations. We then discuss the application of these approaches to phase II trials, including single-arm and randomized phase II trials. A new emerging paradigm of clinical trial design to evaluate different marker-treatment combinations within or across histologically defined cancers in one overarching “master protocol” is also discussed. In phase III trials to confirm clinical utility of a combination of treatment and marker based on a true endpoint (such as overall survival), we discuss enrichment, sequential enrichment, and concurrent, all-comers design approaches. In all-comers, marker-stratified designs, after providing some arugments on the null hypotheses and the level of alpha control (i.e., weak or strong control), we outline various statistical analysis strategies across marker-defined subpopulations. For unstratified all-comers trials, we discuss adaptive design and prospective-retrospective approaches to develop and validate predictive markers or signatures. Lastly, marker-strategy designs for comparison of multiple strategies with regard to the use of new marker are briefly discussed.