We would like to comment on the article by Gottschalk et al,1 which presented evidence for a low-dimensional chaotic process in mood rating of patients with bipolar disorder. If correct, this is an important result for psychiatric research and treatment, because it implies that one could predict and control mood in bipolar patients using models with few variables. The authors' conclusion of low-dimensional chaotic behavior is, however, inconsistent with the results of their power spectral analysis. The following brief discussion may be helpful for those continuing to analyze psychiatric data using nonlinear dynamics.