In the social interaction test of anxiety, microinjections of midazolam (2-8 microg) into the dorsal hippocampus or dorsal raphé nucleus significantly increased the time spent in active social interaction, without changing locomotor activity, thus indicating specific anxiolytic effects. However, tolerance developed to these effects in rats that had been pre-treated for 6 days with (-)-nicotine (0.1 mg/kg/day; subcutaneous). Thus, cross-tolerance to the anxiolytic effects of midazolam develops rapidly following a short period of treatment with a low dose of nicotine, which contrasts with the more slowly developing tolerance (about 3 weeks) that develops after benzodiazepine treatment. Following 6 days of nicotine treatment there was a significant reduction in [(3)H]flunitrazepam binding at 2 and 10 nM in the hippocampus, but no change in the midbrain. The decrease in benzodiazepine binding could explain tolerance to the effects of midazolam when administered to the dorsal hippocampus, but other mechanisms, such as indirect effects on the serotonergic (5-HT) system, might be involved in tolerance to the effects of dorsal raphé nucleus administration.