Four cyclic analogues of the psychoactive phenethylamine derivative 3,4-(methylenedioxy)amphetamine were studied. These congeners, 5,6- and 4,5-(methylenedioxy)-2-aminoindan (3a and 4a, respectively), and 6,7- and 5,6-(methylenedioxy)-2-aminotetralin (3b and 4b, respectively) were tested for stimulus generalization in the two-lever drug-discrimination paradigm. Two groups of rats were trained to discriminate either LSD tartrate (0.08 mg/kg) from saline, or (+/-)-MDMA.HCl (1.75 mg/kg) from saline. In addition, a 2-aminoindan (5a) and 2-aminotetralin (5b) congener of the hallucinogenic amphetamine 1-(2,5-dimethoxy-4- methylphenyl)-2-aminopropane (DOM) were also evaluated. None of the methylenedioxy compounds substituted in LSD-trained rats, while both 3a and 3b fully substituted in MDMA-trained rats. Compounds 4a and 4b did not substitute in MDMA-trained rats. Compounds 5a and 5b did not substitute in MDMA-trained rats, although 5a substituted in LSD-trained rats, but with relatively low potency compared to its open-chain counterpart. In view of the now well-established serotonin neurotoxicity of 3,4-(methylenedioxy)amphetamine and its N-methyl homologue 1, 3a and 3b were evaluated and compared to 1 for similar toxic effects following a single acute dose of 40 mg/kg sc. Sacrifice at 1 week showed that neither 3a nor 3b depressed rat cortical or hippocampal 5-HT or 5-HIAA levels nor were the number of binding sites (Bmax) depressed for [3H]paroxetine. By contrast, and in agreement with other reports, 1 significantly depressed all three indices of neurotoxicity. These results indicate that 3a and 3b have acute behavioral pharmacology similar to 1 but that they lack similar serotonin neurotoxicity.