This study examines clinical predictors of outcome for patients with panic disorder and depression in a 16 week, placebo-controlled trial of alprazolam and imipramine (n = 126). Baseline global severity of illness and phobic avoidance were differentially predictive of acute response to treatment. Patients in the mild to moderate range of global distress experienced smaller degrees of improvement on alprazolam than on imipramine at week 4. At endpoint, the relative effectiveness of the active medication versus placebo was diminished in patients with higher levels of phobic avoidance. This relationship was not evident for completers, suggesting that the adverse effects of avoidance on outcome after sustained treatment was reduced.