The incidence of leukaemia and the mortality from various causes other than cancer observed in the offspring of the Japanese bomb survivors are analysed using linear and exponential forms of a relative risk model. The relative risk coefficients for leukaemia as a function of total pre-conception dose in the offspring of the Japanese and those for children of the Sellafield workforce are compared, and statistically significant differences are found. The statistical significance of these differences is no less marked if attention is restricted to those born before the end of 1950 in the Japanese cohort; therefore it is unlikely that the differences between the pre-conception irradiation leukaemia risks in the Japanese and Sellafield datasets are a result of the different distributions of parental ages at exposure in the two groups, or of the different lengths of time between exposure of spermatogonia and conception. The statistical discrepancies between the two datasets are found to be reasonably robust to possible dosimetric uncertainties in the Sellafield group.