The present study provides a detailed description of a novel drug discrimination procedure employing a T-maze and food reinforcement. Three groups of rats (n = 5) were trained to run one of the two arms of the maze after the i.g. administration of either 1.0, 1.5 or 2.0 g/kg ethanol (chosen as training drug) and the opposite arm after administration of water. Only correct trials were rewarded. All the rats learned the discrimination task. Substitution tests with different doses of ethanol in all the three groups and with the non-competitive NMDA antagonist, dizocilpine (MK-801), in the 2.0 g/kg ethanol-trained group provided a pharmacological validation of the paradigm. Advantages and disadvantages of this procedure are discussed.