Survival from cancer reflects the aggressiveness of the disease, the effectiveness of treatment and host factors such as age. While hospital-based survival rates are typically used to evaluate the care provided in a particular hospital, population-based survival reflects the effectiveness of the overall cancer control strategy in the region. Here, we report the survival experience of 1514 breast cancer patients registered by the Bangalore population-based registry during 1982-1989. There have been very few reports on survival from cancer in India, mainly because of poor patient follow-up and inadequate system of registration of death. This has been largely overcome in this study by means of active follow-up through visits of homes of patients. Scrutiny of medical records and matching with death certificates, was also carried out in a small proportion (12%) of cases. Thus, information on vital status (whether dead or alive) as on January 1, 1993 was available for 1334 (88%) subjects and partial follow up data were available for a further 34 (2%). The observed 5 year survival was 42.3% and the corresponding relative survival was 46.8%. The observed survival was 57.4% for localized disease, 45.8% for direct extension, 37% for those with regional node involvement, 14.2% for distant metastasis and 38.3% for those with unstaged disease. The clinical extent of disease and the educational status were independent predictors of survival.