1. Intracerebroventricular (I.C.V.) injections of isotonic and hypertonic solutions into the dorsal (D3V) and ventral (V3V) third ventricle were employed to examine the release of vasopressin (AVP) and the mean arterial pressure (MAP) response to elevated cerebrospinal fluid (CSF) osmolality in the conscious rat. 2. The D3V injection of hypertonic sodium chloride solution was associated with a concentration-dependent, transient increase in plasma AVP concentration and MAP. 3. The D3V injection of 5 microliters 0.85 M-sodium chloride elicited a 7-fold increase in plasma AVP and oxytocin concentrations, but had no effect on plasma ACTH concentration. The D3V injection of 1.11 M-mannitol in 0.15 M-sodium chloride had no effect on plasma AVP concentration or MAP. However, the D3V injection of 0.746 M-mannitol in 0.4 M-sodium chloride elicited a significant transient increase in plasma AVP, but had no effect on MAP. 4. The V3V injection of 5 microliters 0.85 M-sodium chloride elicited a prolonged increase in plasma AVP concentration and a transient increase in MAP. The V3V injection of 5 microliters 1.11 M-mannitol in 0.15 M-sodium chloride elicited an equal, but transient, increase in plasma AVP concentration, but had no effect on MAP. 5. The pressor effect of a D3V injection of 0.85 M-sodium chloride was unaffected by prior administration of the V1 (pressor) receptor antagonist (beta-mercapto-beta,beta-cyclopentamethylene propionyl1, O-Me-Tyr2, Arg8)-vasopressin. 6. These results indicate that osmotically induced AVP secretion may be mediated by both sodium receptors and osmoreceptors, although expression of the response may depend upon the maintenance of a 'permissive' concentration of sodium in the CSF. 7. It appears also that the pressor effect is not due to increased plasma AVP concentration, but only results from elevation of the CSF sodium concentration.