As part of the cross-national collaborative panic study, a double-blind comparison of alprazolam, imipramine and placebo was performed in Scandinavian outpatients with panic disorder according to DSM-III; 41 patients were randomly allocated to each drug. Doses were increased for 3 weeks to an average of about 6 mg alprazolam, 150 mg imipramine and a corresponding number of placebo capsules, which were then given for 5 weeks. No more than supportive psychotherapy was given. Key symptoms were rated weekly. The drugs were tapered for 4 or 8 weeks and the patients were followed up for 6 months. Compliance at 3 weeks was 95% for alprazolam, 83% for imipramine and 88% for placebo; at 8 weeks 95% for alprazolam, 73% for imipramine and 46% for placebo. At 3 weeks plasma determination showed that the proportion taking diazepam outside the protocol was 0% for alprazolam, 19% for imipramine and 31% for placebo; at 8 weeks the corresponding proportions were 3%, 11% and 16%. Intention-to-treat analysis showed that freedom from panic attacks was obtained for 68% with alprazolam, 61% with imipramine and 34% with placebo. Alprazolam was more effective than imipramine and placebo on anticipatory anxiety and phobic symptoms. Globally rated by physicians and patients, about 60% had complete remission with alprazolam and imipramine and 30% on placebo. At least partial remission was obtained in about 85% with alprazolam, 70% with imipramine and 40% with placebo. Alprazolam had a more rapid onset of action than imipramine on all symptoms. Side effects were generally mild, with a preponderance of drowsiness for alprazolam and anticholinergic effects for imipramine. Tapering was uneventful without significant discontinuation phenomena. During taper and follow-up, several patients in remission relapsed, leaving approximately 30% patients in complete remission in all groups. To obtain more stable improvement, either long-term drug treatment or combinations of drug treatment and psychotherapy should be evaluated.