In March 1981, 40,318 women in Stockholm, aged 40-64, entered a randomized trial of breast cancer screening by single-view mammography alone versus no intervention in a control group of 20,000 women. The attendance rate during the first screening round was 81 per cent and the cancer detection rate was 4.0 per 1000 women. The detection the rate fell to 3.1 per 1000 in the second round, which was completed in October 1985. During 1986 the controlled design of the study was broken and the control women were invited once to screening which was completed the same year. A total of 428 cases of breast cancer were thus diagnosed in the study group and 439 in the adjusted control group. After a mean follow-up of 7.4 years the number of breast cancer deaths in the study and control groups was 39 and 30 respectively. The relative risk of breast cancer death (screening versus control) was 0.71 (95 per cent confidence interval: 0.4-1.2). Among women older than 50 years at entry the relative risk was 0.57 (95 percent confidence interval: 0.3-1.1). Cancer deaths among women under 50 were few and perhaps because of this no mortality reduction was seen in this age group. The estimate of mortality reduction lies between the results from two earlier Swedish randomized controlled trials.