Altered attentional processes for anxiety-linked stimuli are well known. However, it is still unclear whether slower response times in the presence of anxiety-linked stimuli are due to attentional processes, particularly delayed disengagement, or due to freezing, a behavioral response in the face of a threatening situation. Three criteria are necessary to differentiate between delayed disengagement and behavioral freezing: (1) Control of behavioral freezing, (2) control for even distribution of initial focus of attention, and (3) control of the ease with which attention is shifted. We adopted a circular visual search paradigm that met these three criteria. Therefore 80 participants, matched by age and gender, conducted an eye-tracking experiment in two anxiety groups (low vs. high). In the experiment, participants had to saccade away from an irrelevant picture in the center of a screen that could be congruent, incongruent, or neutral to the target. The target was a peripheral anxiety-linked (for high anxiety) picture with a specific feature (direction left or right) to which participants had to respond. We could first replicate previous findings for delayed disengagement for congruent pictures, independent of whether the pictures were anxiety-linked. We further could demonstrate that the high anxiety group compared to the low anxiety group, showed slower manual response times in all conditions. At the same time, they did not differ in disengagement times. Thus, our results indicate that the so-called attentional bias may not be an attentional process but rather a behavioral freezing reaction in the presence of threatening stimuli.