The structure-activity relationships of xanthine derivatives related to the adenosine A(1) receptor antagonists 8-cyclopentyl-1,3-dipropylxanthine (DPCPX) and 1,3-dipropyl-8-(3-noradamantyl)xanthine (KW3902) were investigated by focusing on variations of the 3-substituent. Aromatic residues were well tolerated by the A(1) receptor in that position. A moderate effect of stereochemistry was found for the 3-(1-phenylethyl)-substituted analogue of DPCPX (S>R) at A(1) and A(3) receptors, whereas the opposite stereoselectivity was observed at the A(2) receptor subtypes. A 3-hydroxypropyl substituent was found to be optimal for high A(1) affinity and selectivity. The most potent compound of the present series was 1-butyl-3-(3-hydroxypropyl)-8-(3-noradamantyl)xanthine (10 c), which exhibits a K(i) value of 0.124 nM at rat, and 0.7 nM at human adenosine A(1) receptors, combined with high selectivity (>>200-fold) versus the other receptor subtypes. The similarly potent 8-cyclopentyl-3-(3-hydroxypropyl)-1-propylxanthine was converted into a water-soluble phosphate prodrug, which may become a useful pharmacological tool for in vivo studies. 8-Alkyl-2-(3-noradamantyl)pyrimido[1,2,3-cd]purine-8,10-diones, which can be envisaged as xanthine analogues with a fixed 3-propyl substituent, were identified as a new class of potent, selective adenosine A(1) receptor antagonists. For example, compound 14 (8-butyl-substituted) exhibits a K(i) value of 13.8 nM at human A(1) receptors. A selection of the most potent compounds was investigated in [(35)S]GTPgammaS binding assays and showed inverse agonistic activity. Their efficacy was generally lower than that of the full inverse agonist DPCPX, and depended on subtle structural changes. Some of the new compounds belong to the most potent and selective A(1) antagonists described to date.