AbstractRandom assignment of treatments is an essential feature of experimental design in general and clinical trials in particular. It provides broad comparability of treatment groups and validates the use of statistical methods for the analysis of results. Various devices are available for improving the balance of prognostic factors across treatment groups. Several recentt initiatives to diminish the role of randomization are seen as being potentially misleading. Randomization is entirely compatible with medical ethics in circumstances when the treatment of choice is not clearly identified.