Objective: The study describes breast cancer mortality trends in Tuscany (period 1970–97), comparing Florence with the rest of Tuscany (Florence excluded), and, for Florence, incidence (period 1985–94) and survival (1985–86 versus 1991–92) trends, taking into account the diffusion of screening. Methods: Mortality and incidence rates, age-adjusted on the European population, and 95% confidence intervals (95% CI). Five-year relative survival rates and estimates of risk of dying provided by the Cox model. Results: Mammographic screening, started at the beginning of the 1970s in some municipalities, largely involved the Florence area after 1990 (mammograms/years: from 8000–9000 to 28,000–29,000, respectively, before and after 1990). In the same period no population-based screenings were ongoing in the rest of Tuscany. A significant mortality drop was observed in Tuscany (−3.7%/year), starting at the beginning of the 1990s and observed for ages ≤74 (especially ages 40–49: −11.2%/year). The drop was similar in Florence and in the rest of Tuscany. In ages 50–69, incidence, increasing between 1985–87 and 1988–90 (+6.5%), rose sharply in 1991–94 (+17.0%); it was stable in other ages. Local disease increased more markedly in ages 50–69 (globally: +88.3%), but also in other ages (+20–30%). Regional and metastatic cancers decreased. A significantly better 5-year survival was observed among cases diagnosed in 1991–92, persisting after adjustment by extent of disease. Conclusion: Even if the causes of breast cancer mortality trends are not easy to clarify in an observational study, our data suggest that the drop in mortality observed in Tuscany at the beginning of the 1990s could be largely explained by both earlier detection, outside of an organized screening program, and by better treatments. The increase in incidence and the shift in stage distribution that occurred before the enlargement of the screening area and in age groups not involved in the program, supports the role of a `spontaneous' widespread earlier detection. The better survival of the period 1991–92, only partly explained by the shift in stage at diagnosis, indirectly supports the role of improvement in therapy.