Recent advances in adaptive design methodology have increased consideration of use of such designs in regulatory applications. Two types of adaptation are increasingly seen: sample size re-estimation and adaptive selection. For sample size re-estimation based on a treatment effect estimate at an interim analysis, realization of a possible problem of statistical inefficiency has generated a number of interesting statistical methods. For adaptive selection designs, the concept of using interim data to help selection, e.g., dose, patient subgroup, seems natural and its potentials to advance clinical trial methodology have generate vast interest in exploration space as well as in confirmation space. Despite the desirable advances, some controversies on statistical inference, particularly statistical testing, have recently emerged. This will be the topic of this paper.