Data from fifty-five patients who had hepatic resections for colorectal liver metastases at the National Cancer Institute (NCI) were analyzed to determine the effect of perioperative blood transfusions on disease recurrence and overall survival. Besides blood transfusions, other factors included in the analysis were size, number, and distribution of metastases, margin status of resected metastases, length of disease-free interval, Duke's stage of the primary tumor, type of hepatic resection, and anesthesia time. Using the Cox proportional hazards model, the amount of blood transfused was found to be a significant prognostic factor. For each additional unit of blood transfused the risk of disease recurrence and death was increased by 5% (p = 0.0015) and 7% (p = 0.0013), respectively. The median disease-free survival for patients who received 3-5, 6-10, and greater than or equal to 11 transfused units was 26, 12.1, and 11.4 months, respectively. The median overall survival for patients who received 3-5, 6-10, and greater than or equal to 11 transfused units was greater than 44, 39.2, and 33.6 months, respectively. The number of resected nodules (1-2 vs. greater than or equal to 3), type of resection (anatomic lobectomy vs. wedge resection), and nodule size (less than or equal to 3.0 cm vs. greater than 3.0 cm) were additional factors that were further evaluated to determine the effect of blood transfusions. Analyses stratified for each of these factors revealed that patients who received greater than or equal to 11 units of blood had a significantly decreased disease-free and overall survival compared with patients who received 3-10 units of blood. It is concluded that the amount of perioperative blood transfused is an independent prognostic factor that adversely effects disease-free and overall survival.