Food-aversion learning in pigeons was studied using a concurrent operants methodology. Twelve subjects received extensive baseline training consisting of exposure to a concurrent variableintervall~minvs:nable-intervall-min schedule of reinforcement. A different food, milo or peas, was associated With each schedule. Following the establishment of stable preferences, the birds were given exposure to one of the foods in their home cages. Immediately thereafter, the birds were injected with lithium chloride. For half of the birds, the dosage was 0.3 M; for the other half, it was 0.6 M. Within these groups, for half of the subjects, the injection followed ingestion of the preferred food;for the remaining subjects, it followed the nonpreferred food. Subsequently, the animals were again exposed to the operant choice procedure, and alterations in food preference were assessed. In general, postinjection shifts in preference were greater for birds that received the 0.6 M dosage. In all cases, preinjection levels of preference eventually reappeared. In addition to the effects on food preference, lithium toxicosis also tended to suppress animals' overall rates of keypecking. Taken together, these results support the potential applicability of a concurrent operants methodology to the study of food-aversion learning.