Background <p>Attentional biases to emotional information are assumed to play a crucial role in the onset and maintenance of depression. Moreover, recent studies show that biases may remain present in previously affected individuals during non-symptomatic stages even after acute depression has fully subsided. For example, in an investigation probing attentional disengagement from facial expressions of happiness, sadness, and disgust, never-depressed individuals showed speeded disengagement from disgusted expressions in comparison to happy faces, but this differential processing pattern was absent in currently euthymic individuals with a history of major depression.</p> Purpose <p>Building on these findings, the present follow-up study aimed to explore the predictive power of that previously described disengagement bias by assessing depressive symptoms in 63 initially euthymic individuals six months after they had participated in a gaze-contingent eye tracking task.</p> Methods <p>Each participants’ mean difference in saccade latency to initiate eye movements away from facial expressions of happiness and disgust was assessed at baseline, and tested for associations with self-reported depressive symptom gains six months later.</p> Results <p>The individual’s difference between these two emotion conditions when performing attentional disengagement (ADΔhappiness-disgust) significantly predicted the occurrence of a reliable increase in depressive symptom severity at six months follow-up. This effect remained significant when controlling for baseline symptom severity and lifetime history of depression. Conversely, dimensional change in depressive symptom severity was not predicted by the ADΔhappiness-disgust score.</p> Conclusions <p>We suggest that an individual difference score reflecting the ability to disengage attention from facial expressions of disgust versus happiness may be particularly useful in identifying individuals prone to experiencing reliable increases in depressive symptoms.</p>