Group sequential methods have been applied more often in recent years. By monitoring accumulating response data at regular intervals ("interim analyses"), they provide the ethical and financial advantages of stopping the trial early due to overwhelming efficacy, safety or futility. However, what is the risk of drawing incorrect conclusion based on the small fraction of the data? When and how to design a trial with interim analyses? How to allocate type I (or type II) error (using spending function or customized allocation)? When to use futility boundary? What if there are multiple endpoints, e.g. co-primary endpoints? How to do the estimation after early stopping? Should one use adaptive design instead? These interesting questions and any questions raised by the participants can be discussed at this roundtable.