Pre-treatment predictors of treatment outcome were examined in a group of 144 patients with panic disorder and agoraphobia randomly allocated to alprazolam+exposure (AE), placebo+exposure (PE), alprazolam+relaxation (AR), and placebo+relaxation (PR). First-time psychotropic medication use, severity of agoraphobic disability, and longer duration of illness predicted less global improvement at post-treatment. Pre-treatment severity of agoraphobia predicted less improvement both in the short- and the long-term. Predictors of poorer outcome at 6-month follow-up were older age, past history of depression, severity of phobia targets, and longer duration of illness. Sex, source of referral, pre-treatment depression-anxiety-panic, and expectancy from treatment did not relate to outcome.