To investigate the risks of developing asbestos-related diseases we conducted a historical cohort mortality study on 249 ship repair workers (90 laggers and 159 boiler repairers) in a single U.S. Navy shipyard in Japan. We successfully identified the vital status of 87 (96.7%) laggers and 150 (94.3%) boiler repairers, and, of these, 49 (56.3%) and 65 (43.3%) died, respectively, during the follow-up period from 1947 till the end of 1996. Our in-person interviews with some of the subjects clarified that asbestos exposure was considered to be substantially high in the 1950-60s, decreased thereafter gradually but remained till 1979 in the shipyard. The laggers, who had handled asbestos materials directly, showed a significantly elevated SMR of 2.75 (95% C.I.: 1.08-6.48) for lung cancer. The risk developing the disease was greater in the laggers after a 20-year latency (SMR = 3.42). Pancreatic cancer yielded a greater SMR than unity (7.78, 90% C.I.: 2.07-25.19) in a longer working years group. Four laggers died from asbestosis. The boiler repairers, who had many chances for secondary exposure to asbestos and a few for direct exposure, showed no elevation of the SMR of lung cancer overall, but there was a borderline statistically significant SMR of 2.41 (90% C.I.: 1.05-5.45) in a longer working years group. One boiler repairer died from mesothelioma and four from asbestosis.