To determine what factors predict outcome, the authors retrospectively reviewed the management of all 159 poor-grade patients admitted to Harborview Medical Center at the University of Washington who suffered aneurysmal subarachnoid hemorrhage between 1983 and 1993. Favorable outcome (assessed by the Glasgow Outcome Scale) occurred in 53.9% of Hunt and Hess Grade IV, and 24.1% of Grade V patients. Outcome was largely determined by the initial hemorrhage and subsequent development of intractable intracranial hypertension or cerebral infraction. Using multivariate analysis, the authors developed three models to predict outcome. It was found that predicting outcome based only on clinical and diagnostic criteria present at admission may have resulted in withholding treatment from 30% of the patients who subsequently experienced favorable outcomes. It is concluded that aggressive management including surgical aneurysm obliteration can benefit patients with poor neurological grades and should not be denied solely on the basis of the neurological condition on admission.