OBJECTIVE To examine the occurrence of hypotensive episodes in patients with severe traumatic brain injuries that are not of hypovolemic origin and to investigate possible neurogenic or iatrogenic causes of such episodes. METHODS We reviewed Traumatic Coma Data Bank (TCDB) records of the 248 patients with early hypotension. We attempted to eliminate episodes related to hemorrhagic hypovolemia by excluding patients with (1) extracranial injuries of Abbreviated Injury Scale scores > 3 (n = 99, 40%); (2) postresuscitation hematocrit levels < 35% (n = 76, 30.6%); (3) hematocrit levels decreasing to < 35% during the first 24 hours after injury (n = 47, 19%); and (4) patients with conflicting data (n = 5, 2%). This left 21 patients (8.5%) without discernible extracranial causes for their hypotension. RESULTS Of these 21 patients, 4 had no extracranial injuries and 4 had only a single injury with Abbreviated Injury Scale score = 1. Hypotensive episodes were not associated with terminal or unsalvageable status. Mortality was 43%. Of the multiple factors investigated, the only two that were strongly associated with these "unexplained" hypotensive episodes were the presence of a diffuse injury pattern on computed tomography (n = 15, 71%) and the early use of mannitol or furosemide (n = 16, 76%) (It was policy at TCDB centers that hypotensive patients not receive diuretics until they were resuscitated.) CONCLUSIONS (1) Some episodes of severe traumatic brain injury-related hypotension may be of neurogenic origin. (2) The risk/benefit ratio of early diuretic use in patients with severe traumatic brain injuries may be too high to support liberal use. These data strongly support the need for a study involving prospective collection of data describing the early blood pressure courses in such patients.