Comprehensive review of clinical blood transfusion practice at a tertiary-care medical center is complicated by the extraordinary number of patients that receive such therapy. Computer-assisted review of the key objective data used in making the decisions about transfusion is necessary to evaluate the process. Use of 15,873 units of red blood cells, 3,641 units of plasma, 2,619 pools of platelets or pheresis units, and 259 pools of cryoprecipitate was screened by comparing pre-transfusion and post-transfusion blood counts with the medical staff's evaluation criteria. On this basis, 81.4% of transfusion episodes (TEs) were considered fully justified. Medical records were selected for audit from the cases in which the transfusion decisions could not be justified by on-line information. Abstracted data subsequently justified 82 of 139 audited cases; 68.4% of the comments pertaining to the remaining 57 cases adequately explained the transfusion decision. Thus, nearly 96% of the TEs were justifiable as determined by peer review.