Studies of the frequency of reconstructive surgery among breast cancer patients have involved non-population-based sources, incomplete ascertainment, or both. Postmastectomy breast reconstructive surgery among 4688 breast cancer patients (diagnosed between 1992 and 1996) was estimated by linking a population-based cancer registry with a statewide hospital discharge database in Connecticut. Of these 4688 patients, 585 (12.5 percent) had reconstruction coded in one database or both databases. The reconstruction rates were higher than in a previous study in Connecticut and increased from 9 percent in 1992 to 16 percent in 1996. Reconstruction was not related to the patient's tumor size, marital status, or race (black versus white), but declined with increasing age at diagnosis and with poverty rate (of the census tract of residence). These associations were similar to those reported from previous studies.