In this review, we condense and summarize the results of studies on the therapeutic use of human albumin to promote the more efficient use of this costly resource. Reports of major controlled and uncontrolled therapeutic trials, reviews, and summary articles published in English between 1972 and 1991 were identified through library and MEDLINE searches. Case series, prospective studies, and blinded therapeutic trials were identified from the bibliographies of these sources. All sources were critically evaluated for information about the comparative physiologic results and patient outcomes of the therapeutic use of albumin solutions, crystalloid solutions, and volume expanders other than albumin. The therapeutic use of albumin is of marginal benefit for many conditions for which it has been administered, apparently because of the body's capacity to quickly compensate for rapid colloid osmotic shifts. Human studies show little or no demonstrable value for albumin when it is administered for nutritional supplementation, wound healing, perioperative fluid replacement, treatment of early thermal injury, or therapy during extensive retroperitoneal surgery (including aortic aneurysm resection). Therapeutic albumin has well-defined value in several special circumstances: large-volume paracentesis in cirrhotic patients, acute nephrotic syndromes with diuretic resistance, organ transplantation, and plasmapheresis. Additional studies are needed to compare the efficacy of albumin with other volume expanders. For most purposes, balanced crystalloid solutions are satisfactory substitutes for colloid volume expanders and can be obtained at a fraction of the cost of colloid volume expanders.