Intracerebroventricular (i.c.v.) injection of kassinin produced a prompt and copious drinking response at doses of 10-1000 ng/pigeon, in the absence of other behavioural alterations or of changes in core temperature. Neurokinin A and B evoked drinking, but they were respectively 10 and 100 times less potent than kassinin. Intraperitoneal injection of kassinin elicited drinking, but at doses about 1000 X larger than the i.c.v. ones. The angiotensin antagonist [Sar1, Leu8]angiotensin II did not reduce drinking induced by i.c.v. kassinin, suggesting that its effect is not due to interaction with the central renin-angiotensin system. Moreover, the effect is apparently independent of the mechanisms controlling hypovolaemic and hyperosmotic thirst since exact additivity was found in the dipsogenic response when i.c.v. kassinin was administered in the presence of a hypovolaemic (subcutaneous (s.c.), polyethylene glycol) or hyperosmotic (s.c. hypertonic NaCl) dipsogenic stimulus. The present findings show that kassinin, neurokinin A and B share with the tachykinins already tested (eledoisin, physalaemin, substance P) a common dipsogenic action in pigeons. However, marked differences exist in their dipsogenic potency. This order of potency, eledoisin = kassinin = physalaemin greater than neurokinin A = substance P greater than neurokinin B, is not consistent with the tachykinin receptor subtypes so far proposed.