Seventy panic disorder patients participated in a double-blind, placebo-controlled, randomized infusion study in which sodium lactate and isoproterenol were used to induce panic anxiety. Patients who panicked during lactate, isoproterenol, and placebo infusions generally had higher preinfusion anxiety scores. These findings held true irrespective of the order the infusions were given. Stepwise multiple regression analyses comparing panickers with nonpanickers as the criterion variable revealed that the items "afraid of going crazy," "feeling unsteady," and "feeling paralyzed" on the Panic Description Scale had the highest predicting values.