Researchers studying diet selection of ungulate herbivores have generally considered plant palatability independent of animals' dietary history. However, more recent studies demonstrate that experiences within the life of an animal strongly influence plant selection. We are beginning to understand how food preferences and aversions are formed through gastro-intestinal feedback. My research specifically examines factors that influence the formation of conditioned flavor aversions in the generalist herbivore, sheep. I first examined how variability of food toxicity affects the intake of those foods. I determined that sheep apparently have several mechanisms for regulating intake of toxic foods regardless of whether or not toxic variation can be detected through flavor changes. When changes in flavor correspond to changes in toxicity, animals adjusted intake based on an increase or decrease in toxin concentration. When toxic variation was not detectable through flavor, animals adopted a conservative strategy of eating an amount based on the maximum toxin dose they had experienced.' I was also interested in how illness following the consumption of one food influences the selection of other foods. In diet selection, animals may generalize selection responses among foods with similar flavors. Generalization may be particularly important in the selection of novel foods, i.e., a new food may look, smell, or taste like a familiar food that is preferred or avoided. In several experiments on the generalization of flavor aversions I found that: 1) sheep generalize aversions from familiar to novel foods when both foods had a flavor in common; 2) the more sick an animal got after eating a food the greater the aversion formed to the food and the greater the generalization of that aversion to new foods; 3) the salience or intensity of flavor did not affect the strength of conditioned flavor aversions in sheep on the generalization of the aversion, but this may not always be the case; 4) flavor intensity strongly influenced the acceptance of a novel food. A novel food (wheat) with a strong flavor (3% added ground oregano) was more avoided than a novel food (wheat) with a mild flavor (1% oregano added).