Electrophysiological methods were used to assess the effect of chloride-channel blockers on the macroscopic and microscopic currents of mouse connexin50 (Cx50) and rat connexin46 (Cx46) hemichannels expressed in Xenopus laevis oocytes. Oocytes were voltage-clamped at -50 mV and hemichannel currents (ICx50 or ICx46) were activated by lowering the extracellular Ca2+ concentration ([Ca2+]o) from 5 mM to 10 microM. Ion-replacement experiments suggested that ICx50 is carried primarily (>95%) by monovalent cations (PK : PNa : PCl = 1.0 : 0.74 : 0.05). ICx50 was inhibited by 18beta-glycyrrhetinic acid (apparent Ki, 2 microM), gadolinium (3 microM), flufenamic acid (3 microM), niflumic acid (11 microM), NPPB (15 microM), diphenyl-2-carboxylate (26 microM), and octanol (177 microM). With the exception of octanol, niflumic acid, and diphenyl-2-carboxylate, the above agents also inhibited ICx46. Anthracene-9-carboxylate, furosemide, DIDS, SITS, IAA-94, and tamoxifen had no inhibitory effect on either ICx50 or ICx46. The kinetics of ICx50 inhibition were not altered at widely different [Ca2+]o (10-500 microM), suggesting that drug-hemichannel interaction does not involve the Ca2+ binding site. In excised membrane patches, application of flufenamic acid or octanol to the extracellular surface of Cx50 hemichannels reduced single channel-open probability without altering the single-channel conductance, but application to the cytoplasmic surface had no effect on the channels. We conclude that some chloride-channel blockers inhibit lens-connexin hemichannels by acting on a site accessible only from the extracellular space, and that drug-hemichannel interaction involves a high-affinity site other than the Ca2+ binding site.