The aim of this study was to evaluate the effects by the end of 1999 of the Florence breast screening programme that started in 1990. Approximately 60000 women (aged 50-69 years) were enrolled from 1990 to 1993. Breast cancer cases diagnosed from 1990 to 1996 were partitioned by the method of detection, classified by their tumour size and nodal status and followed-up for mortality at on the 31 December 1999. Incidence-based mortality in the 50-74-year-old women and advanced carcinomas rates were assessed. Due to low compliance (approximately 60%) and the long enrollment phase, only approximately 35% of the total age-specific population person-years were screened. The number of invasive cases diagnosed was 1122, 17% higher than the 958 expected. After the prevalence screening, a reduction of approximately a quarter in advanced carcinomas was observed in the invited women (Odds Ratio (OR): 0.74; 95% Confidence Interval (CI): 0.55-0.98). In the period 1990-1999, 547 breast cancer deaths were observed: 78 (14%) occurred in women invited and half of these in never responders, 385 (70%) occurred in cases diagnosed before screening started. Disproportionate numbers of deaths occurred in women with advanced tumours. The 19% mortality reduction for the invited women was of borderline statistical significance (observed/expected (O/E) deaths: 0.81; 95% CI: 0.64-1.01); by a one-sided test the result would be unequivocally significant. The mortality reduction attributable to screening in the whole population over the 10-year period was 3.2%. The incidence-based mortality analysis confirmed the current follow-up time is too short for screening to have had a major effect on the breast cancer mortality trends. Screening performance might be improved by a higher level of compliance and shorter interval times, but the estimate of the mortality reduction for the invited and the lower rate of advanced carcinomas confirmed that the effect of the programme is in the expected direction.