Résumé : Background: a pilot quality assurance project for breast cancer screening was set up in the Brussels area in 1994. This paper aims to assess the performance of this programme after 4 years of activity, and the specific impact of consensual double reading of mammograms. Methods: each screening mammogram of women aged 50-69 year was submitted to a consensual double reading. Results of readings were registered with standardised forms. Follow-up data were traced for every positive mammogram. Results: 15.624 mammograms were performed in 12.239 women; recall rate at first round was 7,8%, open biopsy rate was 1%, cancer detection rate was 5,8%, positive predictive value of biopsy recommendation was 53,4%, benign to malignant biopsy ratio was 0,87:1, small size (less or equal to 10 mm) cancer proportion was 40%, proportion of cancers free of nodal involvement was 65%. Double reading yielded a 6% gain in sensitivity, while recall rate dropped from 8,1% to 7,8%. Conclusion: apart from a too high recall rate, screening performance was comparable with other published results in the same context; performance indicators ranged within norms recommended by "Europe against cancer". However, impact of double reading was weak and should be re-evaluated in the future perspective of a larger scale organised programme.