A primate model was used to determine whether oxyhemoglobin (OxyHb), methemoglobin (MetHb), or bilirubin is likely to be responsible for cerebral vasospasm following subarachnoid hemorrhage (SAH). Forty cynomolgus monkeys were randomly assigned to one of five groups. On Day 0, each animal underwent angiography followed by right craniectomy and placement of an Ommaya reservoir with its catheter adjacent to the right middle cerebral artery (MCA). The animals received intrathecal injections twice a day for 6 days of one of the following solutions: mock cerebrospinal fluid (CSF); OxyHb; MetHb; bilirubin; or supernatant fluid from an incubated mixture of autologous blood and mock CSF. On Day 7, angiography was repeated and the animals were killed. Comparison of angiograms obtained on Day 0 and Day 7 of the experiment showed significant vasospasm of the right MCA and the right anterior cerebral and internal carotid arteries in the animal groups that had received OxyHb or supernatant fluid. There was a smaller reduction in diameter of the same vessels in the bilirubin group (not statistically significant), while no effects were observed in the groups receiving MetHb or mock CSF. Electron microscopy of the right MCA's gave results consistent with the angiographic findings. One monkey in the OxyHb group developed a delayed-onset right MCA infarction. These data suggest that OxyHb is the cause of cerebral vasospasm following SAH.