The earliest formal sequential procedure, the sequential probability ratio test, involved the plotting of certain test statistics and comparison with straight line parallel boundaries. The boundaries approach can now be used with a wide variety of test statistics, including those appropriate to the analysis of survival data. The boundaries can take various forms, although the use of straight lines still eases the underlying mathematical theory while at least approximating to the requirements of the majority of clinical trials. The implementation of sequential methods needs to be made flexibly and sensitively, with each clinical trial meriting an individualized approach.