1. In cats anaesthetized with chloralose the release of vasopressin in response to nicotine injections was examined. This release was measured by assaying the hormone in samples of venous blood. 2. Nicotine injections were given by three different routes, namely intravertebral, intracarotid and intravenous. The first two represent close arterial routes to the medulla and to the hypothalamus, respectively, the effects of which could be compared to those following intravenous, i.e. systemic, administration. 3. Nicotine was found to increase vasopressin secretion by all three routes of administration. The potency of intracarotid injections was found to be no greater than that of intravenous injections, in sharp contrast to intravertebral injections, which were 4-5 times more potent. 4. In terms of vascular effects, intracarotid and intravenous injections of nicotine were found to increase blood pressure, whereas intravertebral injections of low doses of nicotine were always followed by a fall in blood pressure. Higher doses of intravertebral nicotine produce mixed results, pressor or depressor, in different animals. 5. The vasodepressor effect of intravertebral nicotine was part of a cardiovascular response which included a lowering of total peripheral resistance and of stroke work, whereas the cardiac output, the heart rate and the stroke volume remained essentially unchanged. 6. These results clearly indicate that a medullary area, which has been previously described, is the most sensitive site for the vasopressin releasing action of nicotine and that systemic administration of the drug induces vasopressin secretion by virtue of its action on the medulla, rather than directly on the supraoptic nucleus. 7. The results also indicate that the vasodepressor effect which follows the application of nicotine on the medulla is chiefly due to vasodilator effects on systemic blood vessels, with practically no action on cardiac function. The significance of these results is discussed.