Using low doses of angiotensin II (1-128 ng), a mapping study was carried out to redefine the region within the brain from which the dipsogenic response can be elicited. The most sensitive sites were either close to the anterior cerebral ventricles or were at the tips of cannulae that traversed a ventricular space en route to the tissue site. Conversely, insensitive sites were remote from the ventricles and were not reached by cannulae with a ventricular trajectory. Therefore, a thorough assessment of the role of the ventricular system in the angiotensin thirst phenomenon was demanded. Further studies revealed that direct intraventricular injections were as effective as those made into any tissue site tested. Also, regardless of the similarity in the sites of their termination, cannulae which passed through a ventricular space en route to that site yielded highly sensitive preparations whereas those which did not were insensitive to the hormone. Autoradiographic and radioassay studies showed that tritiated angiotensin II injected into anterior diencephalic tissue through cannulae which traversed the ventricles was rapidly distributed throughout the ventricular system. Thus, efflux of the injected material up the cannula shaft and its entry into the ventricular system is essential for a dipsogenic response to low doses of the hormone. These results support the hypothesis of a periventricular receptor site for angiotensin. It is suggested that systemically generated angiotensin and angiotensin endogenous to the brain may use the ventricular route as a means of access to the sensitive periventricular site.