Long-term outcome (24-month follow-up; 24MFU) of cognitive-behavioral treatment was examined in 63 patients with panic disorder. When the traditional methods of cross-sectional assessment were used (e.g., panic frequency during past month), long-term outcome findings paralleled those of earlier studies. However, assessments of idiographic response and those that examined longer time periods revealed that a large proportion of the sample experienced a fluctuating symptom course of panic-related symptomatology that was not captured by the cross-sectional method. Many (27%) patients sought further treatment for panic during the follow-up period because of a less-than-adequate response to treatment; nevertheless, additional treatment did not result in further clinical improvement. Pretreatment severity of panic disorder was associated with poorer outcome at 24MFU. Use of psychotropic medication during treatment was associated with poorer outcome, although to a weaker degree when pretreatment responding was controlled. Findings are discussed in the context of how the outcome of treatments for panic disorder should be evaluated and interpreted.