To describe the post-transfusion survival of an entire, geographically defined population, we observed all residents of a US county who underwent perioperative transfusion before and after the introduction of a large autologous transfusion programme. We enrolled 444 and 1540 county residents, transfused in 1981 and in 1986-88, respectively. Complete follow-up (until death or for 5 years) was available on 1960 patients (98.8%). Of patients transfused in 1986-88, 67.6% were alive at 5 years, having survived for a mean (+/-SE) period of 46.15 (+/-0.575) months. The survival statistics were 66.2% and 46.09 (+/-1.047) months, respectively, for patients transfused in 1981 (P = 0.8424). Transfusion in 1986-88 vs. 1981 (with 615 [40%] vs. six [1.3%] patients receiving some autologous blood) did not have an effect on survival (P = 0.3892), following adjustment for age, gender, transfusion dose, receipt of a single-unit transfusion and transfusing surgical service. We conclude that the survival of an entire, geographically defined transfused population is substantially longer than that reported previously for patients referred to tertiary-care medical centres. The aggregate 5-year survival of patients transfused in 1981 and in 1986-88 does not differ, despite differences in patient case-mix and in perioperative transfusion practice, particularly as it relates to autologous blood usage.