Toxic epidermal necrolysis is a devastating medication-induced desquamation disorder with a reported mortality rate of 30% to 60% in adults. Data from previously reported series suggest that age, delay in referral to a burn center, total body surface area (TBSA) involvement, and systemic steroid treatment are poor prognostic indicators. We reviewed the records of 39 patients treated in our burn center over the past 10 years and found that the mortality rate was significantly correlated with age, thrombocytopenia, and delay in presentation. Steroid treatment and TBSA involvement were not significantly related to the mortality rate. Thirty-nine adult patients with greater than 20% TBSA epithelial necrosis were cared for in our center from January 1987 to March 1998. Wounds were treated with topical antimicrobial medications and porcine xenografts in a bacteria-controlled nursing unit. We reviewed the records of these patients for 28 clinical characteristics and looked for clinical correlates of mortality by single analysis of variance. The mortality rate was 44% (17 of 39 patients); the cause of death was most commonly multiple-organ dysfunction syndrome, for which a microbial etiologic agent was not always identified. Autopsies were performed on 11 of the 17 patients who died; there was evidence of multiple-organ damage. The patients who survived and the patients who died did not differ significantly in TBSA epithelial necrosis (66%+/-6% vs 72%+/-5%, respectively), admission platelets, number of nosocomial infections, number of complications, preadmission exposure to steroids, or extent of mucosal involvement. When compared with the patients who died, the patients who survived were (1) 20 years younger (47.5+/-4.2 years vs 64.5+/-5.3 years), (2) admitted to the hospital sooner after the onset of their rash (3.5+/-0.4 days vs 5.9+/-1.0 days), (3) much less likely to experience early thrombocytopenia (platelet nadir, 154+/-24 vs 70+/-18), (4) more likely to be febrile on presentation, and (5) less likely to have been treated with antibiotics before referral to our unit. These differences were statistically significant. The most common etiologic agents were antibiotics, anticonvulsants, and nonsteroidal anti-inflammatory drugs. Our results for a group of older patients with toxic epidermal necrolysis with extensive skin involvement suggest that age, delay in hospitalization, thrombocytopenia, and early empiric antibiotic treatment are associated with a poor prognosis.