Fluoxetine is a new antidepressant with pharmacologic effects apparently limited to blockade of neuronal serotonin reuptake. We entered 20 patients who met DSM-III criteria for either panic disorder or agoraphobia with panic attacks into an open, uncontrolled pilot study of fluoxetine. Four responded to placebo in the week before fluoxetine administration and were dropped from the study. Of the remaining 16 patients, nine were nonresponders and seven were responders, with complete cessation of their panic attacks. Eight of the nine nonresponders were unable to tolerate the side effects of fluoxetine. In contrast, all of the responders (and one nonresponder) experienced minimal side effects. Fluoxetine may be effective in the treatment of panic attacks, perhaps implicating the serotonergic system in the pathophysiology of panic disorder. Future studies should use very low doses of fluoxetine to initiate treatment.