Within the overall objective of whether ruminants are able to form conditioned aversions (CFA) toward a food flavour associated with the administration of an aversive stimulus which occurs naturally in food plants (oxalic acid, OA), two specific objectives were tested: (1) whether the rate and degree of formation of CFA are dependent on the dose rate of OA administered and (2) whether the persistence of formed CFA depends on the previous dose rate of OA. Sheep were conditioned to associate the specific flavour of one of two novel foods with either the oral administration of OA or equivalent placebos. Four dose rates of OA were tested (0.06, 0.12, 0.18 and 0.24 g/kg sheep live weight per d), with twelve sheep per dose. Each conditioning period lasted for 8 d and was repeated four times. At the end of each conditioning period the preference for the two flavours was measured in short-term, 20min preference tests. The persistence of the CFA was measured at 0, 7, 21 and 49 d after the completion of the conditioning phase with long-term, 3 h preference tests. The results of the experiment indicated that: (1) the rate and degree of formation of CFA were dependent on the rate of administration of OA; (2) sheep required repeated exposures to the lower dose rates of OA in order to develop CFA and these CFA did not persist in the absence of continual reinforcement; (3) CFA to the higher dose rates of OA were developed after as little as one exposure and persisted over a period of at least 7 weeks. These findings are consistent with the expectation that ruminants should be able to select a diet which minimizes the risk of consumption of potentially harmful foods, whilst at the same time maintaining a degree of flexibility in their feeding behaviour.