One hypothesis of cerebral vasospasm contends that the slow onset and eventual disappearance of the spasm of subarachnoid hemorrhage is caused by the finite production of spasmogens. These are generated and accumulate in the basal cisterns as the result of chemical reactions between blood elements, arterial wall, leptomeninges, and brain. The spasmogens may spread in the general direction of bulk CSF flow to affect arteries more distally. Pharmacodynamic studies performed on isolated cerebral arteries show that a vast array of naturally occurring substances are vasoactive. Among these, derivatives of arachidonate (eicosanoids) are strong candidates as spasmogens because they produce strong, prolonged contractions in human arteries and because the CSF levels of some eicosanoids are preferentially elevated in subarachnoid hemorrhage patients who experience severe spasm. The spasm may be largely refractory to treatment because the intravasation of blood elements and the coagulum serve as barriers to therapeutic agents and because cerebral arteries are devoid of vasa vasorum; however, in animal models of chronic subarachnoid hemorrhage, calcium antagonists given intrathecally will reverse the spasm. Other therapeutic strategems are being tried experimentally and clinically, including compounds like prostacyclin that relax cerebral arteries and inhibit platelets. In vitro studies indicate further that some plasma proteins (e.g., haptoglobin, antithrombin III) and several products of endothelial synthesis (PGI2, EDRF) may naturally provide protection against the genesis of vasospasm. The in vitro responses to bloody CSF are capricious but are not due to many pharmacologically defined substances, including hemoglobin. Hemoglobin and its derivatives, however, may be critical to the enhanced lipid peroxidation and free-radical production that occur in subarachnoid hemorrhage.