Japanese migrants and their offspring on the island of Hawaii and Japanese living in Hiroshima were examined for diabetes mellitus and its vascular complications. the same methods and investigators were used in both locations. Death certificates of Japanese and Caucasians dying on the island during the past 26 yr were analyzed. Diabetes, defined as a venous serum glucose concentration of at least 200 mg/dl 2 h after a 50-g oral glucose load, was significantly more common in the Hawaiian Japanese than in the Hiroshima Japanese subjects. This suggests that diabetes is more prevalent in Japanese in Hawaii than in Japan, although lack of knowledge about the total population of Japanese migrants in Hawaii makes this generalization uncertain. The proportion of deaths attributed to diabetes was much higher in Japanese migrants and their offspring in Hawaii than in Japan. During the 1950s, the proportional death rate from diabetes was about half as large in Japanese Hawaiians as in Caucasian Hawaiians, but it increased to become 1.6 times the Caucasian rate during the 1970s. A nutritional study revealed that the total caloric intake was similar in Japanese in Hawaii and Hiroshima, although the estimated level of physical activity was less in the Hawaiian subjects. Consumption of animal fat and simple carbohydrates (sucrose and fructose) were at least twice as high in Hawaiian as in Hiroshima Japanese. Conversely, Hiroshima Japanese consumed about twice the amount of complex carbohydrate as the Hawaiian Japanese. These observations support the hypothesis that a high fat, high simple carbohydrate, low complex carbohydrate diet and/or reduced levels of physical activity increase risk of diabetes. The proportion of deaths attributed to ischemic heart disease was higher in both diabetic and nondiabetic Japanese Hawaiians than in diabetic subjects in Japan. The rates were similar for Japanese and Caucasians in Hawaii. There was no evidence of an environmental influence on the development of microangiopathy (retinopathy) in diabetes, as the prevalence of diabetic retinopathy (stratified for diabetes duration) was similar in Japanese subjects in Hawaii and in Japan, and it was similar to previous reports from England. On the other hand, diabetes alone did not appear to account for the greater prevalence of macroangiopathy in Hawaiian Japanese than in Hiroshima. Thus environmental factors, possibly including diet, appear to be involved in the development of macrovascular complications of diabetes.