The role of computed tomography (CT) in the management of vasospasm from subarachnoid hemorrhage was evaluated in 242 consecutive cases with CT performed within 7 days after hemorrhage. Only 20% of these cases did not show a detectable subarachnoid hemorrhage on CT. Subsequent angiograms showed vessel narrowing in 56% of the cases; associated clinical deterioration was noted in 34% of the cases. On later CT, clear ischemic areas were detected in 20% of the cases. A strict correlation between the amount of cisternal blood and the subsequent development of vasospasm was observed: although absent or thin cisternal depositions were rarely associated with vasospasm, consistent or thick depositions were frequently linked to vasospasm (72% of the cases) and to ischemic disturbances (51% of the cases), as well as to clear ischemic areas on later CT (30% of the cases). Regarding the morphology of the cisternal blood collection, the risk of developing vasospasm was at its lowest (42%) for depositions only in the frontal interhemispheric fissure and was at its highest (79%) for depositions in multiple cisterns. The site of cisternal deposition corresponded closely to the area of ischemia on later CT. The persistence of subarachnoid blood more than 72 hours after hemorrhage probably increases the risk of vasospasm, although our data are not conclusive. The definition of a CT scan "at risk" for vasospasm--based on the previous findings--gives practical advantages: proper selection of patients in regard to timing of operation, closer observation and the possibility of prophylactic treatment in patients "at risk," and more adequate evaluation of different therapeutic modalities for vasospasm. With regard to the last point, the incidence of vasospasm was not statistically different between two groups of patients uniformly "at risk": the first group submitted to early operation and the second awaiting operation.