Abstract
In designing a clinical trial, the investigator must often decide whether or not to exclude certain patients from randomization. Using Bayesian analysis this paper proposes a criterion for making this decision, based on an extension of Colton's model for the choice between two medical treatments.