We evaluated retrospectively the effect of perioperative blood transfusions on survival in esophageal cancer. The records of all patients who underwent esophageal resection (n = 316) at UCLA Medical Center from 1970 to 1993 were reviewed. Statistical analysis included univariate (log-rank chi 2) and multivariate (Cox proportional hazards) analyses with other known risk factors. High-volume blood transfusions (> 8 units) but not low-volume blood transfusions (1 to 8 units) were associated with a significant decrease in long-term survival (median survival: no transfusion, 22 months; low-volume blood transfusion, 14.5 months, versus high-volume blood transfusions, 6.5 months; p < 0.01). Multivariate analysis revealed that the shorter survival with high-volume blood transfusions was a result of an increased number of postoperative complications. High-volume blood transfusions were not associated with increases in tumor recurrence or infectious complications. The association between shorter survival and high-volume blood transfusions in esophageal cancer may, therefore, be because of the circumstances necessitating transfusion rather than any immunosuppressive effects of the transfused blood. These findings suggest that the transfusion of blood does not by itself decrease the chance of cure after esophageal resection.