A group of patients with severe burns were entered into two sequential prospective randomized trials for reduction of endotoxemia by the use of intravenous polymyxin B. The first group underwent polymyxin administration during the first week after burn injury in a bell-shaped dosage form constructed to resemble the level of endotoxemia as previously documented. This group showed a statistically highly significant reduction in endotoxin levels and a suggestive, but not statistically significant, reduction in wound infection and mortality in the treated group compared with controls. The second group of patients underwent treatment with perioperative polymyxin B given in conjunction with an excisional procedure of the burn wound. In this group, polymyxin B also accomplished a reduction in endotoxemia from preoperative to postoperative cases, but there was no significant reduction in clinical complication rate or mortality. In the dosages used, polymyxin B is nontoxic and promises to be a useful part of the surgeon's armamentarium in dealing with severe complications of gram-negative sepsis.