1. The receptors responsible for 5-hydroxytryptamine (5-HT)-mediated contraction of rabbit isolated epicardial coronary artery denuded of endothelium was examined by bioassay. 2. A variety of 5-HT mimetics caused concentration-dependent contractions. The rank order of agonist potency was 5-carboxamidotryptamine (5-CT) > 5-HT > (+/-)-alpha-methyl-5-hydroxytryptamine ((+/-)-alpha-me-5-HT) = sumatriptan. This was not consistent with relative potencies at any single recognized 5-HT receptor, suggesting the presence of a mixed receptor population. In one subset of preparations precontracted with U46619 (10-30 nM) with the endothelium intact, none of the agonists caused a relaxation. 3. Contractions to 5-HT were antagonized by ketanserin, a 5-HT2A-selective antagonist, but the displacement of concentration-response curves was inconsistent with an interaction between 5-HT and a single receptor population; the slope of regression between antagonist log M concentration and agonist log (concentration-ratio -1) was shallow (0.57). Responses to 5-HT were also antagonized by the 5-HT(1B/1D)-receptor antagonist GR127935 and, again, the slope of regression was shallow (0.68). These data suggest a possible involvement of 5-HT2A and 5-HT1B or 5-HT1D receptors in the response to 5-HT. 4. Contractions to (+/-)-alpha-me-5-HT, which is selective for 5-HT2A over 5-HT1B and 5-HT1D receptors, were competitively antagonized by low concentrations of ketanserin. The regression between antagonist log M concentration and agonist log (concentration-ratio -1) fitted the Schild equation with a slope that was not significantly different from unity (0.95), giving a pA2 value of 9.0. GR127935 (3-30 nM), had no effect on the contractile response to (+/-)-alpha-me-5-HT. These data establish, unequivocally, the presence of 5-HT2A receptors in the tissue. 5. Sumatriptan, a relatively selective 5-HT(1B/1D)-receptor agonist, induced contractions that were antagonized competitively by GR127935 (3-30 nM), although there was a reduction in the maximum response when concentrations of GR127935 exceeded 3 nM. The apparent pA2 (estimated by imposing a unit slope on the log agonist (concentration-ratio -1) value in the presence of 3 nM GR127935) was 8.92. Contractions to sumatriptan were not affected by low (5-HT2A receptor-selective) concentrations of ketanserin, but were antagonized in a competitive manner at higher concentrations (pA2 6.5). These data appear to confirm the presence of 5-HT1B and/or 5-HT1D receptors in the tissue. 6. Antagonism of 5-HT responses by GR127935 was reassessed after blockade of 5-HT2A receptors with 1 microM ketanserin. Under these conditions, GR127935 was able to antagonize 5-HT-induced contractions fully. The slope of regression between log M antagonist concentration and log agonist (concentration-ratio -1) fitted the Schild equation with a slope not significantly different from unity (1.1) (albeit there was still a reduction in maximum response when GR127935 concentration exceeded 3 nM). The apparent pA2 value was 8.8. This reinforces the evidence that 5-HT1B and/or 5-HT1D receptors contribute to the effects of 5-HT in the tissue. 7. In conclusion, in endothelium denuded rabbit epicardial coronary arteries, 5-HT activates 5-HT2A and 5-HT1D and/or 5-HT1B receptors to cause contraction. This appears to be similar to the situation in man.