To evaluate improvement in breast cancer management between 1975 and 2003, we constituted a cohort of 5722 women with an invasive non-metastatic breast cancer. An active follow-up was carried out and completed using administrative databases. Survival rates were computed using a classical person-time and a period analyses. Relative survival rates were also presented. Advanced cancers at diagnosis decreased from 70.7% to 45.5% between 1975 and 2003. Person-time analysis showed a 48.8% reduction in death rates over 30 years. Survival at 10 years ranged from 50% (CI95% = 45-55) for women diagnosed in 1975-1979 to 66% (CI95% = 63-69) for those diagnosed in 1990-1994. Using period analysis, we estimated the 10 year survival of women diagnosed in 2000-2003 to be 68% (CI95% = 64-72). Overall relative survival was 76% (CI95% = 71-80) at 10 years. Taking into account known prognosis factors, survival was significantly improved for recent diagnosis periods (p<0.00001). Observed improvement in survival confirmed 'in real life' data from clinical trials.