Eight dogs were prepared with bilateral carotid loops and trained to stand quietly in a modified Pavlov stand. In each of the dogs, 0.3, 0.45, and 0.6 M NaCl were infused, split between the two carotids at a rate of 0.6 ml-kg-1-min-1 for 10 min and the effect on drinking assessed. A graded increase in drinking was obtained with increased osmolality of the saline (R = 0.59, N = 32, P less than 0.001). Infusion of 0.3 M NaCl was associated with an increase in jugular venous osmolality, but no significant increase in systemic osmolality. Intravenous infusion of 0.3 M NaCl at 0.6 ml-kg-1-min-1 had no effect on drinking. Intracarotid infusion of 0.3 M sucrose in 0.15 M NaCl stimulated drinking to a similar extent as 0.3 M NaCl, whereas intracarotid 0.3 M urea in 0.15 M NaCl had no effect. Increasing the systemic plasma osmolality by infusing 1.04 M NaCl intravenously stimulated drinking, an effect which was abolished by removing the central osmotic stimulus with intracarotid infusions of water. These results are compatible with a central osmoreceptor theory of thirst.