The survival of 1098 patients with ulcer perforation in Norway during the period 1952-1990 was compared with expected survival. Cox regression models incorporating population mortality rates, were used to analyse effects of sex, age, year of birth, and year at risk on excess mortality. Survival was lower in patients than in the general population through a follow-up period of 38 years. Relative survival was lower in women as compared to men, due to more delayed treatment. Long-term survival was lower after praepyloric perforations than after the other perforation types. Relative survival was higher in patients treated 1952-1970 than in those treated more recently. However, adjustment for year of birth revealed a decline in short-term mortality with calendar time, which is in accordance with improved management during the study period. Relative mortality, particularly long-term mortality, was higher in younger birth cohorts, suggesting a shift towards more serious etiologies.