The use of conditioned reinforcers is increasingly promoted in animal training. Surprisingly, the efficiency of their use remains to be demonstrated in horses. This study aimed to determine whether an auditory signal which had previously been associated with a food reward 288 times could be used as a conditioned reinforcer to replace the primary reinforcer in an unrelated operant conditioning procedure. Fourteen horses were divided into two groups of 7: No Reinforcement (NR) and Conditioned Reinforcement (CR). All horses underwent nine sessions of Pavlovian conditioning during which the word “good” was associated with food (32 associations/session). The horses then followed five sessions of operant conditioning (30 trials/session) during which they had to touch a cone signaled by an experimenter to receive a food reward. The last day, horses underwent one test session of the operant response: no reward was given, but the word “good” was said each time a CR horse touched the cone. Nothing was said in the NR group. CR horses did not achieve more correct trials than NR horses during the test. These findings again show that the conditioned reinforcement was ineffective when used instead of the primary reinforcement to maintain conditioning.