Two groups of N/Nih male rats were trained to discriminate saline vehicle from either 2.0 mg/kg (n = 10) or 10.0 mg/kg (n = 10) cocaine in a food-motivated, two-lever operant paradigm. The rats trained at the low-dose cocaine took a significantly longer training period to reach criterion performance than did the high-dose cocaine group. In addition, the ED50 value for the 2.0 mg/kg cocaine-trained animals (0.465 mg/kg) was significantly lower than the ED50 value (2.105 mg/kg) for those animals trained at the 10.0 mg/kg dose of cocaine. This correlation of ED50 values for stimulus generalization decreasing with reduction in training dose was in contrast to the time-course of the two groups when tested from 15 to 240 min post-injection; this experimentation indicated that there was a non-significant difference in half-life for the 2.0 mg/kg (t1/2: 97.1 min) vs. that of the 10.0 mg/kg cocaine-trained group (t1/2: 83.4 min). Generalization tests with other purportedly dopaminergically-active drugs of abuse including 0.05-0.8 mg/kg d-amphetamine, 0.125-1.5 mg/kg methamphetamine and 0.125-1.0 mg/kg methcathinone indicated that the highest doses of each produced generalization and, with the exception of methcathinone, the ED50 values were significantly lower in the low-cocaine trained group. The stimulus properties of cocaine, as they generalize to amphetamine, methamphetamine and methcathinone, can be explained by effects upon central dopaminergic neurons and may be qualitatively different in low-and high-dose trained rats.