Although adenosine is known to activate K+ conduction in atrial tissue, there is still debate as to the involvement of cAMP-dependent mechanisms. In isolated adult guinea pig atrial myocytes, we demonstrate that the highly A1-selective adenosine receptor agonist 2-chloro-N6-cyclopentyladenosine reduced basal cAMP levels by 30-40% in the absence and presence of the nonxanthine phosphodiesterase inhibitor Ro 20-1724. Isoprenaline caused a concentration-dependent increase in cAMP levels, which was more pronounced in the presence of the phosphodiesterase inhibitor. Several adenosine derivatives suppressed the isoprenaline-induced cAMP increase by approximately 80%. The rank order of potency was 2-chloro-N6-cyclopentyladenosine (IC50, 93 nM) greater than (R)-N6-phenylisopropyladenosine (IC50, 309 nM) greater than 5'-N-ethylcarboxamidoadenosine (IC50, 813 nM) much greater than (S)-N6-phenylisopropyladenosine (IC50, 26,300 nM). A similar but complete suppression of the isoprenaline-induced cAMP increase was produced by the muscarinic receptor agonist carbachol (IC50, 398 nM), which like adenosine is known to activate atrial K+ channels. The A1-adenosine receptor-selective antagonist 8-cyclopentyl-1,3-dipropylxanthine antagonized the effect of 2-chloro-N6-cyclopentyladenosine concentration-dependently, with a KB value of 9.6 nM. In atrial myocytes isolated from guinea pigs pretreated with pertussis toxin, the inhibitory effects of adenosine analogs on basal and isoprenaline-stimulated cAMP accumulation were markedly attenuated. It is concluded that the adenosine receptor in guinea pig atrial myocytes, which is known to be linked to K+ channels, is also coupled to adenylate cyclase via a pertussis toxin-sensitive guanine nucleotide-binding protein and shows the characteristics of the A1-adenosine receptor subtype.