In a conditioned taste aversion (CTA) procedure, the consumption of a flavor is followed by the administration of a toxin (e.g. lithium chloride, LiCl), resulting in the future avoidance of the flavor. CTA studies typically make use of forced-exposure paradigms where a volume of the toxin dependent upon the weight of the animal is injected. The use of forced paradigms can be problematic when extended training is required, such as in stimulus discrimination training involving similar flavors, since the animals can be exposed to a high amount of the toxin. In the present study we confirmed the viability of an alternative voluntary-exposure paradigm that more closely mimics natural conditions and is more considerate of the welfare of the animals as a useful tool for investigating discrimination training. In three experiments, rats received free access to either a flavor (sucrose in Experiments 1a and 1b, and saccharin in Experiment 2) or a compound of the flavor mixed with LiCl. The presence of LiCl in the compound induced post-consumption illness. Rats acquired an aversion to the flavor + LiCl compound, thus reducing both their consumption of, and exposure to, LiCl, and gradually increased their consumption of the flavor alone. The present paradigm is more similar to natural conditions than the forced-exposure paradigm as it allows the animals to experience a direct relationship between the amount of the flavor consumed and the magnitude of the illness induced by the toxin.