Recent case-referent studies in the Nijmegen breast-screening programme have shown a reduction in breast-cancer mortality of approximately 50% due to screening of women aged 65 years and older. In this type of study, however, the results may be biased because of self-selection. The purpose of our present study was to compare the breast-cancer mortality rate in a population invited for screening with that of a reference population from an area without a screening programme. In 1977-1978, 6773 women aged 68-83 years were enrolled in the mammographic screening programme in Nijmegen, The Netherlands. The women were followed up until 31 December, 1990. The reference population consisted of women from the same birth cohort from Arnhem, a neighbouring city without mass screening, for whom the entry date was 1 January, 1978. The ratios of the Nijmegen and Arnhem breast-cancer mortality rates with 95% confidence intervals (CI) were calculated. In the study period, 173 patients were diagnosed with primary breast cancer in Nijmegen vs. 183 in Arnhem; 40 Nijmegen patients had died of breast cancer vs. 51 Arnhem patients. The cumulative mortality-rate ratio was 0.80 (95% CI = 0.53-1.22). In the periods 1978-1981, 1982-1985 and 1986-1990, the mortality rate ratios were 1.44 (95% CI = 0.67-3.10), 081 (95% CI = 0.37-1.79) and 0.53 (95% CI = 0.27-1.04), respectively. After adjustment for the difference in incidence rate that existed between the Nijmegen and Arnhem populations, mammographic screening of women older than 65 can be expected to yield a 40% reduction in breast-cancer mortality after 10 years.