Cognitive interventions for bipolar disorder have received increased attention, but very few prospective studies have focused on whether cognitive variables predict the course of bipolar disorder. Available studies have yielded conflicting results concerning the effects of negative cognitions on mania. In this study, 60 individuals with bipolar I disorder completed monthly symptom severity interviews. At a 6-month follow-up, they completed the Dysfunctional Attitudes Scale, the Negative Automatic Thoughts Questionnaire, and the Positive Automatic Thoughts Questionnaire. Participants then completed monthly symptom severity interviews for another 6 months. Cognitive scales were correlated with current depression and predicted increases in depression over time, but were not related to mania. Although results do not support negative cognitions as a predictor of mania, they do provide support for the increasingly common use of cognitive interventions to treat bipolar depression.