Cancer screening is a double-edged sword. On the one hand, screening may offer the opportunity to detect aggressive cancers early - at a point in time when treatment may be more effective - and thereby to reduce subsequent morbidity and mortality. On the other hand, screening may detect harmless cancers that would otherwise never have become apparent and lead to unnecessary treatment. The main aim of this research proposal is to use register-based data from breast cancer screening in Norway, Denmark and S weden to explore the level of overdiagnosis and its causes and consequences. In addition we will explore important sub aims by attacking the four questions: 1) Is there a relative increase in the number of women treated with a) mastectomy, b) lumpectomy, c) radiotherapy, d) tamoxifen and e) adjuvant chemotherapy? 2) Does a simulation model of the relationship between the trends in incidence, clinical staging and mortality of breast cancer in Norway from 1975 to present suggest that screening is associate d with the emergence of tumors with low malignancy potential. 3) Does the cumulated incidence of breast cancers and DCIS in comparable cohorts of women screened biennially during follow-up, or screened only once or twice at the end of follow-up suggest th at overdiagnosed tumors regress spontaneously? 4) Does a nested case-control study of the histopathology of small screen-detected cancers support the proposition that screening preferentially detects clinically insignificant tumors?