1. The cerebrovascular effects of capsaicin have been examined in vitro, in feline isolated cerebral arteries (circular segments, 2-3 mm long, 300-400 microns extended diameter) and, in situ, in pial arterioles (diameter 40-200 microns) on the cortical surface of chloralose-anaesthetized cats. 2. In isolated middle cerebral arteries, low concentrations of capsaicin (10(-14)-10(-10) M) effected a concentration-dependent relaxation of vessels precontracted with prostaglandin F2 alpha. This relaxant response was markedly attenuated by repeated administration of capsaicin but was minimally affected by the presence of atropine, propranolol, cimetidine or spantide in the tissue bath. 3. In isolated middle cerebral arteries, higher concentrations of capsaicin effected a marked concentration-dependent contraction. This contraction was not modified by 10(-6) M phentolamine or 10(-6) M ketanserin. A markedly reduced contraction by capsaicin was found upon the removal of calcium ions from the buffer solution. Also the calcium entry blocker nimodipine reversed the capsaicin-induced contraction. 4. Subarachnoid perivascular microapplication of capsaicin around individual pial arterioles in situ elicited a biphasic response (an immediate vasoconstriction followed by a sustained vasodilatation). The maximum vasoconstriction was a 60 +/- 6% reduction in diameter from base line and the maximum vasodilatation a 38 +/- 7% increase in diameter. Vasodilatation occurred at lower concentrations of capsaicin (EC50, approximately 5 x 10(-8) M) than those required for vasoconstriction (EC50 3 x 10(-7) M). 5. Trigeminal ganglionectomy 10-16 days before the microapplication abolished the in situ vasodilator effects of capsaicin (10(-6) M) applied perivascularly, but was without effect on the vasoconstrictor actions of this agent. 6. Repeated administration of capsaicin (10-6M) around the same arteriole resulted in a progressive attenuation of the vasodilator phase of the response, with no modification of the vasoconstrictor phase. 7. The present study suggests that capsaicin-induced cerebral vasodilatation is due to the release of vasoactive agents from cerebrovascular trigeminal nerve fibres, whereas the vasoconstrictor effect 6f capsaicin is due to a direct effect on the cerebral vasculature which is mediated via the transmembrane passage of extracellular calcium.