Much of the panic literature uses samples that are diagnostically heterogeneous. The panic experience must be homogeneous if we are to combine such heterogeneous groups for research and if we are to use panic characteristics to predict consequences. This study, using the symptom sequences of 21 patients with panic attacks, sought to assess interpatient agreement and compare symptom rankings in patient subgroups. Median polish performed on the symptom sequences suggested interpatient agreement (F = 3.39, p less than or equal to .05). This was further supported by Kendall's coefficient of concordance using symptom runs for various etiologic models (W = 0.338, p less than .0001). The accuracy of the DSM-III to diagnose panic attacks was implied by the lack of association of measures of accuracy--patient effects and diagnostic confidence--with symptom rankings, demographic data, or clinical parameters. In conclusion, this study supports the uniformity of the panic experience.