1. Vasoconstrictor effects of melatonin were examined in isolated rat tail arteries mounted either in an isometric myograph or as cannulated pressurized segments. Melatonin failed by itself to mediate observable responses but preactivation of the arteries with vasopressin (AVP) reliably uncovered vasoconstriction responses to melatonin with maxima about 50% of maximum contraction. Further experiments were conducted with AVP preactivation to 5-10% of the maximum contraction. 2. Responses to melatonin consisted of steady contractions with superimposed oscillations which were large and irregular in isometric but small in isobaric preparations. Nifedipine (0.3 microM) reduced the responses and abolished the oscillations. Charybdotoxin (30 nM) increased the magnitude of the oscillations with no change in the maximum response. 3. Forskolin (0.6 microM) pretreatment increased the responses to melatonin compared to control and sodium nitroprusside (1 microM) treated tissues. The AVP concentration required for preactivation was 10 fold higher than control in both the forskolin and nitroprusside treated groups. 4. In isometrically-mounted arteries treated with nifedipine, melatonin receptor agonists had the potency order 2-iodomelatonin > melatonin > S20098 > GR196429, and the MT2-selective antagonist luzindole antagonized the effects of melatonin with a low pK(B) of 6.1+/-0.1. 5. It is concluded that melatonin elicits contraction of the rat tail artery via an mt1 or mt1-like receptor that couples via inhibition of adenylate cyclase and opening of L-type calcium channels. Calcium channels and charybdotoxin-sensitive K channels may be recruited into the responses via myogenic activation rather than being coupled directly to the melatonin receptors. 6. It is proposed that the requirement of preactivation for overt vasoconstrictor responses to melatonin results from the low effector reserve of the melatonin receptors together with the tail artery having threshold inertia. Potentiative interactions between melatonin and other vasoconstrictor stimuli probably also result from the threshold inertia. A simple model is presented and a general framework for consideration of interactions between weak vasoconstrictor agonists and other vasoconstrictor stimuli is discussed.