Flumazenil (FLU), a benzodiazepine (BZD) partial agonist with a weak intrinsic activity, was previously found unable to precipitate withdrawal in tolerant subjects submitted to long-lasting BZD treatment. The potential use of FLU to treat BZD withdrawal symptoms has also been evaluated tentatively in clinical studies. In the present experiment, FLU (treatment A) was compared with oxazepam tapering (treatment B) and placebo (treatment C) in the control of BZD withdrawal symptoms in three groups of BZD dependent patients. Group A patients (20) received FLU 1 mg twice a day for 8 days, and oxazepam 30 mg in two divided doses (15 mg + 15 mg) during the first night, oxazepam 15 mg during the second night and oxazepam 7.5 mg during the third night. FLU was injected i.v. in saline for 4 hours in the morning and 4 hours in the afternoon, in association with placebo tablets. Group B patients (20) were treated by tapering of oxazepam dosage (from 120 mg) and with saline solution (as placebo) instead of FLU for 8 days. Group C patients (10) received saline instead of FLU and placebo tablets instead of oxazepam for 8 days. FLU immediately reversed BZD effects on balance task and significantly reduced withdrawal symptoms in comparison with oxazepam and placebo on both self-reported and observer-rated withdrawal scales. The partial agonist also reduced craving scores during the detoxification procedure. In addition, during oxazepam tapering, group B patients experienced paradoxical symptoms that were not apparent in FLU patients. Patients treated with FLU showed a significantly lower relapse rates on days 15, 23 and 30 after the detoxification week. Our data provide further evidence of FLUs ability to counteract BZD effects, control BZD withdrawal and normalize BZD receptor function. The effectiveness of FLU may reflect its capacity to upregulate BZD receptors and to reverse the uncoupling between the recognition sites of BZD and GABA, on the GABA A macromolecular complex, that has been reported in tolerant subjects.