The localization of the central cough mechanism was studied by electrically stimulating the lower brainstem in cats lightly anesthetized with pentobarbital sodium or in unanesthetized midcollicular decerebrate preparations. Cough responses were recorded with the aid of a microphone. The cough responsive region was concentrated in an area dorsomedial to the trigeminal tract and nucleus. The sites of action of antitussive agents (dextromethorphan, codeine, clonazepam, diazepam and caramiphen) were studied on the centrally induced cough responses. Each of these drugs was administered intravenously to determine the dorsal levels for cough suppression. In other series, the threshold dose via the intravertebral route was determined first. Because the agent necessary to prevent cough via this route was so small in amount, recovery usually occurred in 20 to 90 minutes. Then the same agent was given intravenously in an attempt to obtain an effective dose which was close to the minimum dose for blocking the cough. The mean effective doses of these agents to abolish the cough via the vertebral artery were only about 1/20 of those required via intravenous injection. The findings suggest that these agents act centrally to suppress the cough responses. Clonazepam was found to be the most potent antitussive among these agents, the mean effective dose being about 1/35 of that of codeine. The antitussive potency of benzodiazepines is not well correlated with their muscle relaxant activity. For instance, clonazepam and diazepam have the same potency in depressing polysynaptic spinal reflexes, whereas the former is 6 times more potent than diazepam as an antitussive. This finding indicates that clonazepam has a high specificity as an antitussive.