A prospective study was performed on 4,262 consecutive patients who had had skull examinations for recent head trauma. Clinical signs and symptoms and patient history were correlated with skull fractures and intracranial sequelae as identified on CT studies, in order to evaluate the predictive value of each clinical finding and to identify high-yield referral criteria. Ninety-seven skull fractures (3%) and 32 intracranial sequelae (0.7%) were observed. All the intracranial complications were observed in patients with fractures and with altered consciousness of some degrees (Glasgow Coma Scale score less than 13). Most patients were asymptomatic (41%) or showed "low risk" symptoms (29%): among them, neither fractures nor complications were observed. High-risk clinical signs, mainly expressing basilar fractures (as rhinorrhea, otorrhea, focal neurologic signs, retroauricular hematoma) demonstrated high predictive value (100%) for intracranial sequelae. Other "moderate risk" findings for intracranial injury--i.e. loss of consciousness at any time, antegrade or retrograde amnesia, multiple trauma, and possible skull penetration--showed a high correlation with skull fractures and a slightly lower one with intracranial sequelae. The most predictive finding for brain injury was the depressed level of consciousness: brain injuries were never observed in fully conscious patients; in altered consciousness with GCS 15-13 we observed 4% of skull fractures with no sequelae; at GCS values 12-9, 61% of skull fractures and 20% of sequelae were present, whereas at GCS less than 8, 100% of complicated fracture were observed. The finding of skull fracture showed 33% of predictivity for brain damage, which was, however, always associated with "high or moderate risk" clinical signs. Therefore, the authors suggest some guidelines for the management of patients with recent head trauma, including referral criteria for X-rays or CT studies, based on signs and symptoms with high, intermediate and low risk of developing intracranial sequelae.