When ethanol is used as a training stimulus in drug discrimination experiments, benzodiazepines (such as diazepam) as well as non-competitive N-methyl-D-aspartate (NMDA) antagonists (such as ketamine) substitute for ethanol; in contrast, when a benzodiazepine or an NMDA antagonist is used as a training drug, ethanol does not substitute reliably. In the present experiments, we trained rats to discriminate a mixture of diazepam and ketamine, to test the hypothesis that ethanol would substitute for this drug combination. Using a two-lever choice procedure with food as a reinforcer, 22 rats were trained to discriminate a mixture of diazepam (5.6 mg/kg) and ketamine (10 mg/kg) from vehicle. When administered as a mixture, diazepam and ketamine substituted for the training mixture in a dose-dependent manner. When administered separately, diazepam or ketamine substituted for the mixture with full substitution occurring at 5.6 and 17.8 mg/kg, respectively. Ethanol almost completely substituted for the mixture at 1 g/kg. There was no cross-substitution between diazepam and ketamine in rats trained to discriminate diazepam (5.6 mg/kg, n = 10) or ketamine (10 mg/kg, n = 12) from vehicle. In addition, ethanol did not substitute for the training drug in either of these discriminations. These results suggest that the simultaneous action of GABAA agonist and NMDA antagonist mechanisms produce a greater ethanol-specific discriminative stimulus than activation of either component individually.