Analysis carried out here generalized on earlier studies of chromosomal aberrations in the populations of Hiroshima and Nagasaki, by allowing extrabinomial variation in aberrant cell counts corresponding to within-subject correlations in cell aberrations. Strong within-subject correlations were detected with corresponding standard errors for the average number of aberrant cells that were often substantially larger than was previously assumed. The extrabinomial variation is accommodated in the analysis in the present report, as described in the section on dose-response models, by using a beta-binomial (beta-B) variance structure. It is emphasized that we have generally satisfactory agreement between the observed and the beta-B fitted frequencies by city-dose category. The chromosomal aberration data considered here are not extensive enough to allow a precise discrimination between competing dose-response models.