We recently reported the identification of a novel human adenosine A3receptor-selective agonist, (2 S,3 S,4 R,5 R)-3-amino-5-{6-[5-chloro-2-(3-methylisoxazol-5-ylmethoxy)benzylamino]purin-9-yl}-4-hydroxytetrahydrofuran-2-carboxylic acid methylamide (CP-608,039), with 1,260-fold selectivity for the human A3versus human A1receptor (DeNinno et al., J Med Chem 46: 353–355, 2003). However, because the modest (20-fold) rabbit A3receptor selectivity of CP-608,039 precludes demonstration of A3-mediated cardioprotection in rabbit models, we identified another member of this class, (2 S,3 S,4 R,5 R)-3-amino-5-[6-(2,5-dichlorobenzylamino)purin-9-yl]-4-hydroxytetrahydrofuran-2-carboxylic acid methylamide (CP-532,903), which both retained human A3receptor selectivity (210-fold; human A3/human A1Ki: 23/4,800 nM) and had improved rabbit A3receptor selectivity (90-fold; rabbit A3/rabbit A1Ki: 23/2,000 nM). Infarct size was measured in Langendorff hearts or in vivo after 30 min of regional ischemia and 120 min of reperfusion. Five-minute perfusion with CP-532,903 before ischemia-reperfusion elicited a concentration-dependent reduction in infarct size in isolated hearts (EC50: 0.97 nM; maximum reduction in infarct size: 77%, P < 0.05 vs. control). Furthermore, administration of CP-532,903 (150 nM) at reperfusion also significantly reduced infarct size by 64% ( P < 0.05 vs. control), which was not different ( P ≥ 0.05) from the cardioprotection provided by the same concentration of drug given before ischemia. The selective rabbit A1receptor antagonist BWA1433 did not affect CP-532,903-dependent cardioprotection. In vivo, CP-532,903 (1 mg/kg) reduced infarct size by 50% in the absence of significant hemodynamic effects (mean arterial pressure, heart rate, rate-pressure product). CP-532,903 and CP-608,039 represent a novel class of human A3receptor-selective agonists that may prove suitable for investigation of the clinical cardioprotective efficacy of A3receptor activation.