OBJECTIVETo review the natural history and determine the rates of intra- and extralesional hemorrhaging of brainstem cavernous malformations (cavernomas) monitored by one neuro-ophthalmology service.METHODSA record review of all patients with brainstem cavernomas who were evaluated by a neuro-ophthalmology service between 1987 and 1999 was performed. We recorded the clinical symptoms and Rankin disability grade at presentation, during the worst clinical episode, and at the last follow-up examination. Magnetic resonance imaging scans were reviewed for evidence of intralesional hemorrhage (a bleeding episode), edema, or venous anomalies, and the cavernoma size was assessed.RESULTSThirty-seven patients (age range, 6–73 yr; mean age at presentation, 37.5 yr) underwent a mean of 4.9 years of follow-up monitoring. At presentation, there were 27 bleeding events and 8 nonhemorrhagic events; 2 patients did not exhibit symptoms. Patients who were at least 35 years of age exhibited a lower risk of bleeding episodes (odds ratio, 0.15; 95% confidence interval, 0.1–0.4). Cavernomas of at least 10 mm were associated with a higher risk of bleeding episodes (odds ratio, 3.48; 95% confidence interval, 1.3–9.4). Thirty-nine bleeding episodes occurred in 31 patients, yielding a bleeding rate of 2.46%/yr. There were eight rebleeding episodes, yielding a rebleeding rate of 5.1%/yr. Three patients experienced extralesional bleeding episodes; all of these patients experienced rebleeding. Of the 39 follow-up magnetic resonance imaging scans, the cavernoma size was unchanged in 66.7%, smaller in 18%, and larger in 15%. At the last follow-up examination, the mean Rankin grade was 1.0 for all patients, 0.6 for the 25 nonsurgically treated patients, and 1.4 for the 12 surgically treated patients.CONCLUSIONRebleeding is not more common among patients who first present with bleeding, and it often has little effect on the neurological status of patients. Significant morbidity attributable to a brainstem cavernoma occurred in 8% of patients during follow-up monitoring of medium duration.