To evaluate the possibility that abnormal regulation of benzodiazepine receptor (BZR) function may relate to the pathophysiology of anxiety disorders, two doses of the BZR antagonist flumazenil (Ro 15-1788) and placebo were administered to panic disorder patients in a double-blind, randomized, crossover design. All 11 patients received flumazenil (600 mg). Ten of the 11 also received flumazenil (200 mg), and 8 of the 11 were given matching placebo capsules orally on separate test days. Neither dose of flumazenil showed significant anxiolytic effects or significantly altered heart rate, blood pressure, or levels of plasma cortisol or 3-methoxy-4-hydroxyphenylglycol in comparison to placebo. On 200 mg days, anxiety ratings on a visual analog scale significantly increased compared to placebo at 30 min and then returned to baseline levels. Panic attacks occurred on placebo, 200 mg, and 600 mg days in 0/8, 4/10, and 0/11, respectively. Failure to observe anxiolytic effects suggests that flumazenil at these doses does not antagonize a tonic increased interaction of BZRs with an endogenous BZR inverse agonist in panic patients.