In 3 experiments rats experienced 2 flavors, each paired with sucrose, in order to establish a conditioned preference to each. One (flavor Fe) was then presented alone (an extinction procedure) prior to a choice test between Fe and the flavor that did not undergo extinction (Fne). Hungry rats showed a preference for Fne over Fe (Experiment 1A), but rats that were not food-deprived showed no effect of extinction when given a choice between Fe and Fne immediately after extinction (Experiment 1B) or after an interval in which reexposure to sucrose was given (Experiment 2). The extinction procedure was not without effect, however, as Fe was preferred over Fne after sucrose had been devalued by pairing with lithium chloride, and Fne was preferred over Fe after a procedure likely to enhance the value of the sucrose (Experiment 3). The explanations considered propose that preference conditioning establishes a range of associations between the flavor and the various properties of sucrose (its nutritional value, its taste, the hedonic reaction it evokes). It is suggested that the form of learning that mediates revaluation effects is sensitive to extinction whereas that responsible for performance on a consumption test is not.