Abstract
The apparent failure of the BEIR absolute risk model to explain the data on the Japanese atomic-bomb survivors does not imply that the BEIR relative risk model (RRM) is correct. RRM is objectionable in that it fits the data only in conjunction with an assumption not in accord with current knowledge and thinking. Contrary to what is widely believed, RRM is not a consequence of, or consistent with, initiator-promoter theories; models derived from initiator-promoter theories fit the data with fewer adjustable parameters and without requiring unpalatable assumptions. The preferable models give substantially lower radiation risks.