Cognitive biases in information processing play an important role in the etiology and maintenance of emotional disorders. A new methodology to measure attentional biases is presented; this approach encourages subjects to scan and re-scan images with different thematic content, while the pattern of their attentional deployment is continuously monitored by an eye-tracking system. Measures of attentional bias are the total fixation time and the average glance duration on images belonging to a particular theme. Results showed that subjects with depressive disorder (n=8; Beck Depression Inventory Score>/=16) spent significantly more time looking at images with dysphoric themes than subjects in the control group (n=9). Correlation analysis revealed that the differences between the fixation times of the two groups are significantly correlated with the valence ratings, but not with the arousal ratings of the images. The average glance duration on images with social, neutral and threatening themes were similar for both groups, while the average glance duration on images with dysphoric themes was significantly larger for subjects with depressive disorder. The above results suggest that subjects with depressive disorder selectively attend to mood-congruent material and that depression appears to influence the elaborative stages of processing when dysphoric images are viewed.