[en] The recent reassessment (DS86) of radiation doses at Hiroshima and Nagasaki will have a profound influence on radiation protection standards. One important aspect is the almost order of magnitude reduction in estimated neutron doses at Hiroshima: in the previous dosimetry, the generally increased (per dose) hazard at Hiroshima compared to Nagasaki was interpreted as being due to neutrons having a greater per dose effectiveness, and, on this basis, RBEs were estimated. For example, the RBE for all cancers except leukemia could be estimated to be between about 15 and 50 (80% confidence level). However, the corresponding estimate with the new dosimetry is between 0 and 90 (50% confidence level). Thus it appears, prima facie, that no useful RBE data can be derived from the epidemiological data at Hiroshima. An attempt is made in this study to generate RBEs at Hiroshima based on radiobiological considerations, but making as few assumptions as possible