STUDY OBJECTIVE: To describe time trends in gastric cancer incidence in Zaragoza and Navarre, and to investigate time period and birth cohort as determinants of such trends. DESIGN: Cases from two registries were grouped into five year intervals and the following were calculated: age specific and sex specific incidence rates, and the male to female ratio. Log linear models including age, period of diagnosis, and birth cohort were fitted. SETTING: The Zaragoza Cancer Registry covers the province of Zaragoza, which has a population of 824,776 (403,755 men and 421,021 women). The Navarre Cancer Registry covers the province of Navarre which has 512,512 inhabitants (254,786 men and 257,726 women). In both cases population figures were based on the late census. PATIENTS: These comprised incident cases of gastric cancer reported to the Zaragoza Cancer Registry in 1963-87 and to the Navarre Cancer Registry in 1973-87. MAIN RESULTS: Navarre registered higher adjusted and cumulative rates than Zaragoza for both sexes. In both provinces, there were relative declines in the rates for men and women of 3% and 4% respectively per year. In Zaragoza, the risk of developing stomach cancer fell in generations born between 1888 and 1933, and rose in subsequent birth cohorts in both sexes, while in Navarre the cohort effect showed an approximately linear risk for both sexes. Both provinces recorded increases in risk associated with cohorts born between 1933 and 1943. CONCLUSION: The incidence rates of gastric cancer fell in both Zaragoza and Navarre. The reason for the greater incidence of gastric cancer in Navarre remains unknown. Trends in rates seem to be mainly linked to birth cohort. Increases in risk in generations born after 1933 may be ascribable to nutritional deficiencies in the early years of life.