We are very pleased to have received the comments of Doctors Barnett and Edelmann regarding the random selection of patients for placement in a study. We would heartily concur that random, rather than alternate or serial allocation, would avoid unconscious bias of a cyclic nature which might run in phase with the allocation cycle. It is also possible that if the investigators were aware of the treatment group to which the patient would be allocated, an additional bias would be introduced which could clearly affect selection of patients for study.