Long-term treatment leads to tolerance to and dependence on benzodiazepines. Abrupt termination of benzodiazepine administration triggers the expression of signs of dependence. Mice withdrawn from chronic treatment with diazepam showed a time-related evolution of anxiety, muscle rigidity, and seizures between days 4 and 21 after treatment discontinuation. A period between withdrawal days 1 and 3 was symptom-free. Surprisingly, during this "silent phase" the susceptibility of mice to alpha-amino-3-hydroxy-5-tert-butyl-4-isoxazolepropionate (ATPA) and kainate seizures and the magnitude of monosynaptic reflexes mediated by non-N-methyl-D-aspartate (NMDA) mechanisms were enhanced. In apparent contrast, the "active phase", between withdrawal days 4 and 21, was characterized by increased susceptibility to NMDA seizures and enhanced magnitude of polysynaptic reflexes, which are NMDA dependent. Treatment of mice with alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionate (AMPA) antagonists 1-(4-aminophenyl)-4-methyl-7,8-methylenedioxy-5H-2,3-benzodiazepine (GYKI 52466) or 2,3-dihydroxy-6-nitro-7-sulfamoylbenzo(f)quinoxaline but not with the NMDA antagonist 3-[(+/-)-2-carboxypiperazin-4-yl]-propyl-1-phosphonate (CPP) during the silent phase prevented signs of dependence. In contrast, treatment with CPP but not with GYKI 52466 during the active phase prevented the symptoms. The development of tolerance to and dependence on diazepam was prevented by concurrent treatment of mice with CPP but was not prevented by GYKI 52466. These data indicate that NMDA-dependent mechanisms contribute to the development of tolerance to diazepam and to the expression of signs of dependence in mice after termination of long-term treatment with diazepam. Nevertheless, the non-NMDA-mediated silent phase is essential for triggering the symptoms. Therefore, AMPA antagonists may offer a therapeutic approach for preventing dependence on benzodiazepines that is an alternative to NMDA antagonism.