Background. In a controlled trial comparing acyclovir with vidarabine in the treatment of neonatal herpes simplex virus (HSV) infection, we found no significant difference between the treatments in adjusted mortality and morbidity. Hence, we sought to define for the entire cohort (n = 202) the clinical characteristics that best predicted the eventual outcome in these neonates. Methods. Data were gathered prospectively at 27 centers between 1981 and 1988 in infants less than one month of age who had virologically confirmed HSV infection. We examined the outcomes by multivariate analyses of 24 variables. Disease was classified in one of three categories based on the extent of the involvement at entry into the trial: infection confined to skin, eyes, or mouth; encephalitis; or disseminated infection. Results and Conclusions. Results and Conclusions. There were no deaths among the 85 infants with localized HSV infection. The mortality rate was significantly higher in the 46 neonates with disseminated infection (57 percent) than in the 71 with encephalitis (15 percent). In addition, the risk of death was increased in neonates who were in or near coma at entry (relative risk, 5.2), had disseminated intravascular coagulopathy (relative risk, 3.8), or were premature (relative risk, 3.7). In babies with disseminated disease, HSV pneumonitis was also associated with greater mortality (relative risk, 3.6). In the survivors, morbidity was most frequent in infants with encephalitis (relative risk, 4.4), disseminated infection (relative risk, 2.1), seizures (relative risk, 3.0), or infection with HSV type 2 (relative risk, 4.9). With HSV infection limited to the skin, eyes, or mouth, the presence of three or more recurrences of vesicles was associated with an increased risk of neurologic impairment as compared with two or fewer recurrences. (N Engl J Med 1991;324:450–4.). © 1991, Massachusetts Medical Society. All rights reserved.