To investigate the relationship between angiotensin II (ANG II) and mean arterial pressure (MAP) in the control of drinking in rats, we infused ANG II intravenously at constant rates (either 50 or 100 ng.kg-1.min-1 for 90 min) and varied MAP by intravenous injections of diazoxide (5-20 mg/kg). Rats were pretreated with captopril to block the endogenous synthesis of ANG II. When given alone, low and high doses of ANG II increased MAP approximately 30 and 50 mmHg, respectively. The low but not the high dose significantly increased water intake above control levels. Both doses caused such a large diuresis and natriuresis that the net effect was fluid loss. Reducing MAP toward normal greatly increased the drinking response to the high but not the low dose of ANG II and reduced the urinary solute and water loss to both doses. These results support the hypothesis that water intake and net fluid gain are inhibited when MAP is above normal. When MAP was reduced below normal in rats given constant infusions of ANG II the amount of water drunk and net fluid gain was proportional to the dose of ANG II but not the dose of diazoxide, the degree of hypotension, or urinary losses. This is consistent with previous reports that ANG II is essential for the drinking response to hypotension. Furthermore, it demonstrates that ANG II is not merely permissive but probably the signal controlling water intake when arterial pressure is reduced below normal.