To test the safety of a large intrathecal dose of human recombinant tissue plasminogen activator (rt-PA), 6 cynomolgous monkeys were given 10 mg of rt-PA (mean, 2.7 mg/kg) through an Ommaya reservoir after craniectomy and dissection of the basal cisterns. Bleeding occurred briefly at the incision in 2 animals; otherwise the clinical condition of all 6 remained normal throughout the postoperative period. Systemic fibrinolysis did not occur, and gross and microscopic examination of the brain and meninges revealed no abnormality. Next, we evaluated the efficacy of unilateral administration of rt-PA suspension (0.5 mg) plus slow-release gel rt-PA (1.25 mg) in lysing a bilateral subarachnoid clot and preventing vasospasm in a randomized, placebo-controlled trial. Sixteen monkeys were divided randomly into 2 equal groups, each of which underwent baseline cerebral angiography, followed by frontotemporal craniectomy and induction of subarachnoid hemorrhage on the left side and then on the right. Before closure on the right side either rt-PA or an equal volume of placebo was injected into the subarachnoid space. On day 7 angiography was repeated, and the animals were killed under anesthesia for necropsy. One of the animals in the placebo group developed a cerebral infarction on day 5. In the placebo group significant vasospasm occurred in all major right and left-sided anterior cerebral vessels (P less than 0.01). No vasospasm occurred in the rt-PA-treated animals. Whereas gross subarachnoid clot was found in all animals in the placebo group (mean clot weight 1.13 g), only a small fragment of clot was found in a single rt-PA-treated animal.(ABSTRACT TRUNCATED AT 250 WORDS)