In separate, parallel experiments, rats were allowed to sham or real feed a flavored sweetened condensed-milk solution after a 17.5-h fast. Coinciding with administration of the milk were injections of insulin or vehicle (saline). Insulin had no effect on intake in the real feeding situation, and flavors paired with insulin (0.8, 2.5, or 5.0 U/rat) were avoided, relative to saline-paired flavors. In sham-feeding rats, insulin (0.8 U/rat) significantly reduced the amount of milk consumed, and flavors paired with insulin injection were preferred, relative to saline-paired ones. Insulin produced a relatively larger hypoglycemia during sham feeding than during real feeding. Thus it appears that insulin-paired flavors of milk may be preferred by animals when they are ingested in sham feeding but not in animals real feeding. Perhaps rapid gastric emptying, as would occur with a liquid diet, accompanied by increased insulin levels, leads to malaise and flavor aversion.