This study investigated the effects of introducing random reinforcement training after a drug discrimination between 3.0 and 15.0 mg/kg chlordiazepoxide had been acquired in rats; a two-lever food-rewarded operant procedure was used. Matched on the basis of dose-generalization test data, two dose-equiisensitive groups were formed ( A and B). Group A received 30 daily saline injections, which were followed by a random reinforcement session during which either left or right lever presses were reinforced on a probabilistic basis on each of the trials comprising a session. Group B received saline but no training or testing during this period. Subsequent testing revealed that responding conditioned to the low-dose training condition, but not to the high-dose training condition, was significantly changed in Group A. The randon reinforcement data further suggested that Group A did not discriminate saline from the low training dose. In Experiment 2, Group B of Experiment 1 was submitted to 30 random reinforcement training sessions, each preceded by a high training dose administration. Data showed that responding to the high dose, but not the low dose, was changed. For both experiments, chlordiazepoxide dose generalization following reacquisition was similar to that obtained before the random reinforcement phase. The findings indicate that the response pattern changed only for the stimulus condition present during random reinforcement. The random reinforcement manipulation did not disrupt the original discrimination between the high- and low-dose conditions.