Background In France, as in other countries, breast cancer care has changed due to therapeutic advances and organized screening programs. Can the effect of new therapeutic procedures over time be measured by health care institutions considering these changes ? The present study used data from a hospital-based cancer registry to analyze changes in 5-year disease-specific and event-free survival among women with primary breast cancer over three time periods (1990-1993, 1994-1997, and 1998-2001). Methods All cases of primary invasive breast carcinomas, initially treated in a French Comprehensive Cancer Center between 1990 and 2001, were included. In situ breast carcinoma and male breast cancer were excluded. Cox proportional hazards models were used to analyze disease-specific and event-free survival (DSS and EFS) rates over the three time periods (1990-1993, 1994-1997, and 1998-2001). Results During the 1990–2001 period, 4,165 primary breast cancers were initially treated at the Comprehensive Cancer Center. Out of 1,012 deaths overall, 74.6% were due specifically to primary breast cancer (respectively 98% from cancer itself and 2% from treatment side effects); the cause was unknown for only 3.3% of deaths. Out of 3,810 complete remissions, 18.2% presented local, regional or metastatic relapse and 3.8% presented a second primary breast cancer. Comparison of DSS and EFS rates in a recent reporting period (1998–2001) with those in earlier time periods (1994–1997 and 1990–1993) indicated that substantial survival gains were achieved with respectively 88.4% (95% CI: 86.4–90.5), 83.2% (95% CI: 81.3–85.2), and 79.8% (95% CI: 77.4–82.2) (p < 0.01) for 5-year Disease-Specific Survival, and respectively 78.3% (95% CI: 75.7–81.0), 73.9% (95% CI: 71.6–76.3), and 70.1% (95% CI: 67.4–72.8) (p < 0.01) for 5-year Event-Free Survival. After adjustment for prognostic factors, period was identified as an independent predictor of survival. Conclusion Survival improvement is likely to be due to changes in routine clinical practice such as an increased use of systemic adjuvant therapy over the study periods, dose modification of epirubicin in adjuvant chemotherapy for node-positive breast cancer since 1994, and organized screening programs since 1997. However the effect of possible early diagnosis and over-diagnosis biases due to screening cannot be assessed.