Design: The ability of performance on the emotional Stroop task to predict subsequent levels of symptomatology was examined within a longitudinal design in a group of individuals with SAD. Participants were tested in the winter when depressed and again in the summer when remitted. Results: There was no change in emotional Stroop performance across the two time points. However, winter performance on the emotional Stroop task predicted symptom levels in the summer, even with winter symptom levels partialled out. The nature of this relationship was that a greater emotional Stroop effect for threat words in the winter was related to more improved mood in the summer. Furthermore, the strength of the relationship between lower winter emotional Stroop performance and more negative winter mood was associated with poorer mood in the summer.