BACKGROUND AND PURPOSE: A simple clinical scale of intracerebral hemorrhage (ICH), comprising the Glasgow Coma Scale score, age, infratentorial origin, ICH volume, and intraventricular hemorrhage, was recently shown to predict 30-day mortality. We studied how well the original ICH Score would predict morbidity and mortality and determined whether modification would improve the predictions. METHODS: Patients admitted to a regional hospital with acute ICH in 1999 were reviewed. Independent predictors of mortality or good outcome (modified Rankin score <or=2) at 30 days were identified by logistic regression to devise a new ICH Score for comparison with the original Score. A modified Score was created by substituting National Institutes of Health Stroke Scale (NIHSS) for the Glasgow Coma Scale. RESULTS: The mortality rate was 22%, and 35% had good outcome. Independent factors for mortality were high NIHSS score, intraventricular hemorrhage, subarachnoid extension, and narrow pulse pressure. Independent factors for good outcome were low NIHSS score and low admission temperature. For all ICH Scores, no patient had a maximum score of 6. Cutoff values of >or=3 and <3 provided the best Youden's index of diagnostic test in all ICH Scores for mortality and good outcome, respectively. The original and modified ICH Scores predict mortality equally well. The new and modified ICH Scores are slightly better for prediction of good outcome. CONCLUSIONS: All 3 ICH Scores are simple clinical grading scales. As reliable predictors of good outcome and/or mortality, they are useful in clinical research studies and standardization of clinical protocols.