Characterization of A2B receptors is hampered by the lack of selective pharmacological probes and often relies on their relative affinity to agonists that are selective at other receptor types. This approach is limited because the affinity of A2B receptors for putative A3 agonists has not been determined. Using the human erythroleukemia cell line HEL as a cellular model for A2B-mediated adenylate cyclase activation, we found the following potencies (pD2) for the non-selective agonist 5'-N-ethylcarboxamidoadenosine (NECA) (5.65 +/- 0.04), the putative A3 agonists N6-benzyl-NECA (4.17 +/- 0.06) and N6-(3-iodobenzyl)-N-methyl-5'-carbamoyladenosine (IB-MECA) (3.7 +/- 0.02), and the A2A agonist 4-[(N-ethyl-5'-carbamoyladenos-2-yl)-aminoethyl]-phenylpropionic acid (CGS21680) (2.8 +/- 0.1). Because of the lack of a selective agonist, characterization of A2B receptor function is difficult in cells co-expressing A2A receptors. In the human mast cell line HMC-1, NECA induced cAMP accumulation with a concentration-response relationship best fitted to a two-sited model (pD2 7.69 +/- 0.42 and 5.92 +/- 0.21 for high- and low-affinity sites), suggesting the presence of both A2A and A2B receptors in these cells. We demonstrated that A2B receptors can be selectively activated with NECA in the presence of the selective A2A antagonist 5-amino-7-(phenylethyl)-2-(2-furyl)-pyrazolo[4,3-e]-1,2,4-triazolo[1,5-c ]pyrimidine (SCH 58261). Under these conditions, the concentration-response relationship of NECA for cyclic AMP accumulation was now best fitted to a one-site model (pD2 5.68 +/- 0.03, Hill slope 0.93 +/- 0.06, 95% confidence intervals 0.8 to 1.06) corresponding to selective activation of A2B receptors. Using the approaches developed in this study, we determined that A2B, and not A2A or A3, receptors account for all the calcium mobilization induced by NECA in HMC-1 cells.