Atomic bomb survivors are a population suitable for studying the relationship between somatic mutation and cancer risk because their exposure doses are relatively well known and their dose responses in terms of cancer risk have also been thoroughly studied. An analysis has been made of erythrocyte glycophorin A (GPA) gene mutations in 1,226 atomic bomb survivors in Hiroshima and Nagasaki. The GPA mutation frequency (Mf) increased slightly but significantly with age at the time of measurement and with the number of cigarettes smoked. After adjustment for the effect of smoking, the Mf was significantly higher in males than in females and higher in Hiroshima than in Nagasaki. All of these characteristics of the background GPA Mf were in accord with those of solid tumor incidence obtained from an earlier epidemiological study of A-bomb survivors. Analysis of the dose effect on Mf revealed the doubling dose to be about 1.20 Sv and the minimum dose for detection of a significant increase to be about 0.24 Sv. No significant dose effect for difference in sex, city, or age at the time of bombing was observed. Interestingly, the doubling dose for the GPA Mf approximated that for solid cancer incidence (1.59 Sv). And the minimum dose for detection was not inconsistent with the data for solid cancer incidence. The dose effect was significantly higher in those diagnosed with cancer before or after measurement than in those without a history of cancer. These findings are consistent with the hypothesis that somatic mutations are the main cause of excess cancer risk from radiation exposure.