We conducted two experiments to determine how toxicosis affected preference of sheep for foods varying in flavors, nutrients, and a toxin. The first experiment determined how toxicosis affected the preference of lambs (fed a basal ration of alfalfa pellets) for foods that varied in energy and a toxin. Thirty lambs (10/treatment) were given LiCl by gavage (0, 50, or 100 mg/kg body wt/day), and 1 hr later were offered for 15 min/day foods containing different amounts (low, medium, high) of energy (barley) and a toxin (LiCl) added to alfalfa. The proportions of barley and LiCl changed every three to six days during the 30-day study. The results showed: (1) lambs' food preferences were high > medium > low for barley in the absence of LiCl; (2) lambs quickly regulated intake of foods in response to changes in barley and LiCl concentrations, even with short exposures (15 min/day); (3) lambs maintained intake of LiCl at about 57 mg/kg body wt by adjusting intake of food containing LiCl in accord with the amount of LiCl they received by gavage; and (4) as barley levels increased, intake of foods containing LiCl increased. The second experiment determined the relative influence of flavors, nutrients, and toxins on food preferences of lambs. We did this by treatments in which different flavors (onion and oregano at 1%) were paired with different levels of energy (depending on the addition of wheat to rabbit pellets) or a toxin (LiCl). At six-day intervals, we varied the types of food offered, either changing the nutrient or toxin content and the flavors. The resulting analyses of preference showed lambs markedly preferred foods high in nutrients and low in toxins, regardless of flavor, when changes in food flavor were not correlated with changes in nutrient and toxin concentrations. Thus, in both experiments lambs quickly regulated intake of foods varying in nutrients and a toxin according to the lambs' toxicological and nutritional state. Even with brief eating bouts lambs discriminated accurately and exhibited little permanent preference or aversion in postconditioning preference tests. The lambs remained in an unbiased testing mode, sampling anew the food. This is adaptive because the toxin and nutrient contents of plants vary with season and location. Most taste aversion studies emphasize the permanence of aversions and miss the dynamic sampling power of animals.