Recent uncontrolled research suggested that trazodone and sodium valproate may be helpful in benzodiazepine (BZ) discontinuation. We therefore undertook a double-blind study to assess whether trazodone and valproate, as compared to placebo, would attenuate withdrawal and facilitate discontinuation in BZ-dependent patients with a minimum of 1 year daily BZ use. Seventy-eight patients, taking a mean dose of 19+/-17 mg/day of diazepam (or its equivalent), were stabilized for several weeks on their BZ (16 diazepam, 25 lorazepam, 37 alprazolam) and then for 1-2 weeks, pretreated with trazodone, sodium valproate or placebo before being tapered at 25% per week. All treatments were continued for 5 weeks post-taper. BZ-free status was assessed after 5 and 12 weeks post-taper. Neither trazodone nor valproate had any significant effect on withdrawal severity. Peak physician withdrawal checklist change from baseline to peak severity was 16.4 for trazodone, 18.04 sodium valproate and 18.24 placebo (F = 0.10; NS). Taper success rates were significantly effected by both active agents at the 5-week, but not 12-week, assessment. At 5 weeks post-taper, 79% of sodium valproate and 67% of trazodone, but only 31% of placebo patients were BZ-free (chi2 = 7.34; df 2; P<0.03). Major adverse events for trazodone were sedation and dry mouth, and for valproate, diarrhea, nausea and headaches.