The question as to whether a threshold dose exists in radiation carcinogenesis is important for the estimation of radiation risk at low doses. Although the relationship between the radiation dose and tumor incidence fits a multihit or linear-quadratic model, a threshold model can not be excluded. In fact, the radiation dose at which no increase in tumor incidence is recognized over the background level can be seen in the induction of solid tumors1) and leukemia2-5) with single whole body exposure to X-rays or gamma rays, as well as in pulmonary tumors6) caused by partial irradiation with single doses of up to 2.5 Gy of X rays in mice. When irradiation is protracted or the total dose fractionated, no apparent increases in tumor incidence in skin cancer in mice and rats occurs7-9), or in ovarian tumor10) and leukemia in mice5,11,12). Errorfree repair of DNA damaged by a low dose, low dose rate, or fractionated exposure is considered to be efficient. We established the conditions of irradiation that yield 100% cumulative tumor incidence by repeated beta-ray radiation to the backs of mice 3 times a week13,14). I here summarize our results which show the threshold effect in radiation carcinogenesis under our conditions and the associated p53 gene mutations.