The Cambridge Structural Database (CSD) was searched through two 3D queries based on substructures shared by well-known antagonists at the A(1) and A(3) adenosine receptors (ARs). Among the resulting 557 hits found in the CSD, we selected five compounds to purchase, synthesize, or translate synthetically into analogues better tailored to interact with the biological targets. Binding experiments using human ARs showed that four out of five tested compounds turned out to be antagonists at the A(1)AR or A(3)AR with K(i) values between 50 and 440 nM. Lead optimizations of 2-(benzimidazol-2-yl)quinoxalines (BIQs, 3) gave the best results in terms of potency and selectivity at the A(1) and A(3) ARs. Particularly, 2-(4-ethylthiobenzimidazol-2-yl)quinoxaline (3e) exhibited K(i) values at the A(1)AR, A(2A)AR, and A(3)AR of 0.5, 3440, and 955 nM, respectively, whereas 2-(4-methylbenzimidazol-2-yl)quinoxaline (3b) displayed at the same ARs K(i) values of 8000, 833, and 26 nM, respectively.