We discovered that a food aversion could be conditioned in anesthetized sheep. Sheep were allowed to eat a familiar food (alfalfa-grain pellets) for 30 min, and 90 min later they were given either an intraruminal (IR) injection of water (C), an IR injection of LiCl (L), anesthesia followed by an IR injection of water (A), or anesthesia followed by an IR injection of LiCl (A+L). Induction of anesthesia was by an intraveneous injection of pentobarbitone sodium, and maintenance of deep anesthesia was by halothane. Sheep were maintained in deep anesthesia for 2 h to ensure that the effects of LiCl on the acquisition of a food aversion, which occur within about 1 h, were completed before they awakened. When tested 5 days later, sheep that received LiCl (treatments L and A+L) consumed less alfalfa-grain pellets than sheep that did not receive LiCl (treatments C and A) (241 g vs. 306 g; p = 0.057). Intake of sheep that were anesthetized (treatments A and A+L) did not differ from that of sheep that were not anesthetized (treatments C and L) (295 g vs. 252 g; p = 0.183). Nor was there an interaction between LiCl and anesthesia (p = 0.423). Thus, we conclude that changes in preferences for foods caused by postingestive feedback occur automatically every time food is ingested (i.e., they are noncognitive), and the kind and amount of feedback is a function of the match between the food's chemical characteristics and its ability to meet the animal's current demands for nutrients.