A dynamic treatment regime (DTR) is a sequence of decision rules, one per treatment decision, for choosing effective treatments for individual patients that can adapt over time to an evolving illness. Instead of delivering the same type and dosage of treatment to every patient, a DTR assigns different treatment types/dosages across patients and within each patient across time according to the patient’s values on tailoring variables. And the goal is to determine the best treatment at each time point to maximize long-term clinical outcome. There has been some discussion regarding the design and the inference of DTR in the past. However, there still are open questions in practice. For example, for the randomization-based inference, there might be re-randomization of the patient entry sequence involved. When there is not enough sample size, we might not be able to have enough configurations and p-value might be too conservative. It is of interest to investigate whether the traditional analyses (under any specific conditions) are valid for this case. This session is to discuss the possible implementation of the traditional methods with DTR, as well as several new statistical learning tools, including Q-learning and to introduce the techniques for designing phase II and phase III clinical trials for discovery and verification of individualized treatment regimens.