A biennial mammography screening programme started for all women aged 50-69 in the municipality of Copenhagen, Denmark, on 1 April 1991. We report here on the outcome of the first three invitation rounds. Copenhagen has approximately 40,000 female residents aged 50-69. The Copenhagen Municipality IT Service runs the data files and issues the invitations based on the daily updated population register. Screening takes place at a special clinic at Bispebjerg Hospital, and since 1 September 1996 all assessment and surgery take place at Rigs-hospitalet. The first invitation round covered the period April 1991 to April 1993, the second invitation round the period May 1993 to May 1995, and the third invitation round the period June 1995 to March 1997. By now in total 120,772 invitations have been issued, 84,036 screening tests have been performed, 4110 women have been assessed with additional tests, 1057 women have undergone surgery, and 697 cases of invasive breast cancer cases or carcinoma in situ have been detected. The number of women screened during each of the three invitation rounds as a percentage of the target population aged 50-69 was 71%, 65% and 63%, respectively. It is possible for a woman to notify the programme if she does not want to be invited, and this caused the coverage to decline. The participation rate among the invited women was 71%, 69% and 70%, respectively. Ninety percent of those screened the first and second times and invited the third time participated. The number of screen detected invasive breast cancers of carcinoma in situ cases per 1000 participants during each invitation round was 11.9; 6.3; and 6.1, respectively. When compared with the incidence of invasive breast cancer in Copenhagen before screening, the ratios were 4-7 for all first time screened and 2 for all second or third time screened. Among the 697 screen detected cases, 84% were invasive breast cancers and 11% carcinoma in situ in women not previously known with breast cancer, and 5% were second primary breast cancers. Fifty two invasive interval cancers were found when women who tested negative during the first invitation round were followed up for two years in the Danish Cancer Register and the files of invasive breast cancers of the Danish Breast Cancer Cooperative Group. Compared with the incidence of invasive breast cancer in Copenhagen before screening this gave a proportionate interval cancer rate of 0.34 (95% confidence interval 0.26-0.45). Eight percent of the women who ever participated in the programme had experienced at least one false positive screening test, and 10% of the women who participated all three times in the programme had experienced at least one false positive test. The assessment was an efficient procedure for the sorting out of the false positive screening tests. During the third invitation round, three quarters of those who underwent surgery had either invasive breast cancer or carcinoma in situ. The participation rate in the present programme is relatively low as are the participation rates in other large cities. However, the programme is well accepted among those who participate, as a high proportion of these women come back. The detection rates compared with the previous incidence of invasive breast cancer indicate a good performance of the programme, and so does the proportionate interval cancer rate following the first invitation round. The programme is thus functioning well. The relatively low coverage and the flow in the target population limit the reduction in breast cancer mortality to be expected from the screening programme in the Copenhagen municipality.