We have identified subtype selective agonists, partial agonists and antagonists, which distinguish the human recombinant Mel1a and Mel1b melatonin receptors expressed in COS-7 cells. Melatonin receptor agonists showed higher affinity for competition of 2-[125I]-iodomelatonin binding for the Mel1b than the Mel1a melatonin receptor. The dissociation constants (Ki) of 16 agonists determined on the recombinant human Mel1a and Mel1b melatonin receptor subtypes showed a significant correlation (r2 = 0.85, slope = 0.97, P < 0.0001, n = 16). However, six agonists showed 10 to 60 fold higher affinity for the Mel1b melatonin receptor as indicated by the affinity selectivity ratios (Mel1a/Mel1b) [8-methoxy-2-acetamidotetraline (11); S20098 (14); 8-methoxy-2-propionamidotetraline (20); 6, 7 di-chloro-2-methylmelatonin (21); 6-chloromelatonin (57); 6-methoxymelatonin (59)]. Dissociation constants for competition of 11 partial agonists and antagonist for 2-[125I]-iodomelatonin binding were between 15.5 (luzindole, pKi: 7.7) to 362 (4-phenyl-2-chloroacetamidotetraline, pKi: 9.1) fold higher for the Mel1b than for the Mel1a melatonin receptor. The lack of correlation between the pKi values (r2 = 0.23, P > 0.1, n = 11) strongly suggest that the two human melatonin receptor subtypes can be distinguished pharmacologically. The partial agonist: 5-methoxyluzindole (pKi: 9.6) and the competitive melatonin receptor antagonists: GR128107 (pKi: 9.6), 4-phenyl-2-chloroacetamidotetraline (pKi: 9.1), 4-phenyl-2-acetamidotetraline (pKi: 8.9) and 4-phenyl-2-propionamidotetraline (pKi: 8.8) are selective Mel1b melatonin receptor analogues as their affinity selectivity ratios (Mel1a/Mel1b) are bigger than 100. We conclude that the 40% overall amino acid difference in the sequence of the human recombinant Mel1a and Mel1b melatonin receptors is reflected in distinct pharmacological profiles for the subtypes. We compared the pharmacological profile of the presynaptic ML1 melatonin heteroreceptor of rabbit retina mediating inhibition of the calcium-dependent release of dopamine to that of the recombinant Mel1a and Mel1b melatonin receptors. Melatonin inhibited [3H]dopamine release by 50% (1C50) at 20 pM with a maximal inhibitory effect (80%) at 1 nM. The partial agonists, i.e., N-acetyltryptamine (1C50 5.6, maximal inhibition 55%) and 5-methoxyluzindole (1C50: 1.3, maximal inhibition 40%) showed various degrees of efficacy while none of the competitive melatonin receptor antagonists did inhibit [3H]dopamine release on their own. The potency (1C50) of full melatonin receptor agonists significantly correlated with their affinity to compete for 2-[125I]-iodomelatonin binding to either the Mel1a (r2 = 0.76, slope = 0.77, P < 0.0001, n = 17) or Mel1b (r2 = 0.63, slope = 0.75, P < 0.001, n = 17) human melatonin receptors. By contrast, the apparent dissociation constants (KB) for partial agonists and antagonists to antagonize the inhibition of [3H]dopamine release mediated by activation of the ML1 heteroreceptor by melatonin, significantly correlated with the affinity constants (Ki) for 2-[125I]-iodomelatonin binding determined of the Mel1b (r2 = 0.77, slope = 0.55, P < 0.001; n = 11) but not the Mel1a (r2 = 0.27, P < 0.1, n = 11) subtype. Together these results demonstrate that the pharmacological profile of the human recombinant Mel1b melatonin receptor is similar to that of the functional presynaptic melatonin heteroreceptor of rabbit retina, which we referred as an ML1B subtype. We conclude that the selective Mel1b melatonin partial agonists and antagonists described here can be used to identify melatonin receptor subtypes in native tissues and to search for subtype selective analogues with therapeutic potential.