Italian population-based breast cancer screening programmes with 2-year, high-quality mammography started in the cities of Florence and Turin in the early 1990s. Breast cancer cases from the local Tumour Registry were classified by method of detection and tumour characteristics (size, nodal-status and grade). Follow-up was at December 2001. In total, 4444 breast cancer cases were analysed. The Hazard Ratio comparing before and after-invitation breast cancer cases indicated a 27% reduction (HR=0.73; 95%CI: 0.61-0.87) in the risk of dying for the disease. After adjustment for tumour characteristics, survival rate was comparable by invitation status, whereas the proportion of early cancer was 33.7% and 46.6% in the before and after-invitation group. Survival rates by tumour characteristic subgroups was comparable by invitation status. Late stage and grade 3 were indicators of poor prognosis. Adjustment for tumour characteristics confirmed screening and not differential treatment as the most important reason for the observed survival benefit. The survival analysis by specific subgroups did not support the hypothesis that the difference in prognosis was attributable to differential treatment.