Ventriculocisternal perfusion (VCP) or localized "push-pull" perfusion (PPP) was performed in pentobarbital-anesthetized dogs. Renal function was studied in protocols consisting of a 1-h experimental period bracketed by 1-h control and recovery periods. The animals were perfused with artificial cerebral spinal fluid (CSF) throughout the protocol but the sodium concentration was elevated (300 mM, high Na) during the experimental period. In the first series of experiments, prior to VCP, petroleum jelly plugs were injected into either the ventral third ventricle (V3V) or other brain regions. When petroleum jelly plugs were injected into the V3V prior to VCP, they prevented perfusates from reaching this area, as determined by subsequent dye perfusion. These V3V plugs blocked the natriuresis and ADH response otherwise induced by high Na VCP. In the second series of experiments, the anterior portion of the V3V (AV3V) was perfused by means of PPP. In some of the PPP experiments a V3V plug was also injected to ensure restriction of the stimulus to the AV3V. PPP of the AV3V with high Na CSF at 50 microliter/min did not induce a natriuresis (either with or without a V3V plug). However, when the high Na CSF was perfused at a very high rate (100 microliter/min) and the stimulus was not restricted to the AV3V by a V3V plug, a natriuresis did occur. These results suggest that the ventricular surface where high Na CSF acts to cause natriuresis and increased ADH secretion is within the V3V region but not in the AV3V.