OBJECTIVE: To evaluate the effectiveness of screening for breast cancer as a public health policy. DESIGN: Follow up in 1987-92 of Finnish women invited to join the screening programme in 1987-9 and of the control women (balanced by age and matched by municipality of residence), who were not invited to the service screening. SETTING: Finland. SUBJECTS: Of the Finnish women born in 1927-39, 89893 women invited for screening and 68862 controls were followed; 1584 breast cancers were diagnosed. MAIN OUTCOME MEASURES: Rate ratio of deaths from breast cancer among the women invited for screening to deaths among those not invited. RESULTS: There were 385 deaths from breast cancer, of which 127 were among the 1584 incident cases in 1987-92. The rate ratio of death was 0.76 (95% confidence interval 0.53 to 1.09). The effect was larger and significant (0.56; 0.33 to 0.95) among women aged under 56 years at entry. 20 cancers were prevented (one death prevented per 10000 screens). CONCLUSIONS: A breast screening programme can achieve a similar effect on mortality as achieved by the trials for breast cancer screening. However, it may be difficult to justify a screening programme as a public health policy on the basis of the mortality reduction only. Whether to run a screening programme as a public health policy also depends on its effects on the quality of life of the target population and what the resources would be used for if screening was not done. Given all the different dimensions in the effect, mammography based breast screening is probably justifiable as a public health policy.