Objective:To assess whether clinical variables might be useful in selecting patients who will have an acute intracranial abnormality seen in head computed tomographic scans (HCT).Design:Retrospective study.Setting:Medical intensive care unit (MICU) in a tertiary teaching hospital.Measurements:Medical records of patients admitted to the MICU who underwent HCT between January 1, 1994, and December 31, 1995, were reviewed. Patients with acute intracranial abnormalities (HCT-positive) and those without new acute findings (HCT-negative) were compared on various clinical variables, including demographics, indications for obtaining the HCT (mental status change, neurologic deficit, fever, seizures), coagulation profiles, when the HCT was performed (at admission or after admission), and ordering physician.Main Results:Of 297 HCTs obtained in 230 patients, 37% (109/297) were positive. When the clinical variables were examined univariately, only the presence of a neurologic deficit (70% vs. 37%; difference, 33%;p< .001) differed significantly between positive and negative HCTs. Multivariate analysis confirmed that only the frequency of a new neurologic deficit differed significantly in the two groups (p< .001; odds ratio, 3.9; 95% confidence interval, 2.3-6.4). In patients without neurologic deficits, only the presence of seizures was associated with a positive HCT (p< .01: logistic regression). The presence of either neurologic deficit or seizures best predicted a positive HCT: sensitivity 0.81, specificity 0.53, positive predictive value 0.50, and negative predictive value 0.83.Conclusion:Among MICU patients, the presence of either neurologic deficit or seizures is associated with the presence of an acute intracranial abnormality seen in HCT, but the association is not powerful enough to reliably depend on these clinical variables to select patients for HCTs in the MICU.