Drinking after intragastric hypertonic solutions was examined to determine whether increased plasma osmolality always accompanied initiation of drinking. A 2-ml infusion through a gastric catheter was the beginning of tests in Sprague-Dawley male rats. Latency to drink was shorter and 1-h water intake was greater for increasing concentrations of NaCl (600, 1,200, and 1,800 mosmol/kg) compared with baseline (290 mosmol/kg). Although 600, 900, or 1,200 mosmol/kg NaCl elicited drinking, such infusions failed to change systemic plasma osmolality, and 900 mosmol/kg also failed to change plasma sodium, protein, renin activity, or packed cell volume at the initiation of drinking. Intragastric 900 mosmol/kg sodium bicarbonate, sodium isethionate, potassium chloride, lithium chloride, and mannitol differentially increased water intake. Total subdiaphragmatic vagotomy abolished drinking elicited by intragastric NaCl; selective gastric or hepatic vagotomy attenuated intake under some conditions. These results support the hypothesis of a vagally mediated, gastrointestinal and/or hepatic-portal, osmosensitive mechanism for initiation of drinking in advance of postprandial increases in systemic osmolality.