Analytical epidemiologic studies are less effective when dietary habits are homogenous within a population. For that reason, we performed chronological and spatial correlation analyses between mortality from stomach and large intestinal cancers and various foods/nutrients intakes. The age-adjusted death rates (AADRs) for male colon cancer were strongly associated with the population size and those for female colon cancer and male rectal cancer were weakly associated with the population size, but the AADRs for stomach cancer in both sexes and those for female rectal cancer were not associated. From 1969-71 to 1981-83, the AADRs for stomach and female rectal cancers decreased and the AADRs for colon and male rectal cancers increased. With an increasing population size and from the earlier period to the recent period, the intakes of western-style fat-rich foods such as butter & margarine, cheese and ham & sausage increased and those of rice, fish and some traditional Japanese foods decreased. The results of geographical correlation analyses were generally consistent with the results from the urban-rural variations. The chronological correlation analyses suggested that stomach cancer might be associated with Japanese-style dietary habits after almost no lag time and colon cancer might be associated with westernized dietary habits after a lag time of about 10 years.