The history of total body irradiation (TBI) is surprisingly old. Low-dose TBI was routinely employed for leukemia before the era of chemotherapy. Following the brilliant success of Thomas et al. in the 1970s, bone marrow transplantation (BMT) appeared to be the sole curative treatment modality for high-risk leukemia. In addition, a supralethal dose of TBI was widely accepted as a form of preparation for BMT. The records of 365 patients who underwent BMT between 1975 and 1985 were collected from 31 hospitals by the IVth national survey conducted in Japan. Of these, 264 patients (74%) were classified as having leukemia. As of September 1986, 157 of these leukemia patients had died, interstitial pneumonitis (IP) being the leading cause of death (32%). Using Cox's proportional hazard regression model, it was indicated that the status of the disease and clinical condition at the time of BMT as well as supportive treatment had a great impact on survival after BMT for acute leukemia. On the basis of the clinical condition at the time of BMT, 2-year survivals were 57% and 16% for patients in remission without infection and the remaining patients, respectively (p = 0.0001). Nonremission at the time of BMT (p = 0.0138), advanced age (p = 0.0115) and increased number of platelet transfusions (p = 0.0351) were found to be significant risk factors associated with IP, while TBI was also one of the most important factors in the development of IP. Two-year probabilities of developing IP were 81% and 49% in single-dose TBI and in fractionated TBI, respectively (p = 0.0002). In the single-dose TBI group, a dose rate of less than 5 cGy/min resulted in a low incidence of IP. In the fractionated TBI group, the incidence of IP was somewhat low in the 6-fraction group. According to my own experience, 4-field TBI of 12 Gy/6 frx/3 days with a carefully monitored lung dose of 8 Gy in total resulted in significant improvement in BMT for leukemia.