AbstractThis paper reviews aspects of the development of sequential analysis of clinical trial data in medicine and suggests simple strategies for progress. The emphasis is on the pragmatic and ethical requirements of aspects of the design of phase III trials and in circumstances of genuine uncertainty characterized by much clinical experimentation. In particular consideration is given to the consequences of determining sample sizes from incorrect estimates of treatment effects. Armitage's work on sequential trials is traced to simple group sequential procedures based on repeated significance tests to minimize expected sample sizes in a wide class of experimental situations.