A large body of evidence suggests that processing of affective information is typically disrupted in anxiety. It has also been hypothesized that anxious individuals are less able to evaluate contextual cues and to respond in an adaptive way to stress. In the present study, 25 participants (16 females; 9 males) scoring high (scores of 45 or above) and 26 participants (13 females; 13 males) scoring low (scores of 35 and below) on a standardized measure of trait anxiety performed an emotion search task to investigate attentional biases when the task provides an explicit emotional context. An emotional context was set in each block by asking participants to look as quickly as possible at a face expressing a specific emotion, while eye movements were being recorded. On each trial, two faces appeared, one of them expressing the target emotion and the other one expressing a distractor emotion. High trait-anxious participants showed slower response times (time to look at the instructed emotion), regardless of the affective context, compared to the control group. Additionally, we found slower responses to happy faces (positive context) in the anxious group in the presence of neutral and fearful distractors. Cognitive control may therefore be disrupted in anxiety, as anxious people take longer to process (search for) happy faces, presumably because attentional resources are drawn by neutral and fearful distractors. Those differences were not observed in a simple reaction times task, which suggests that attentional biases, and not differential processing of low-level facial features, are responsible for those differences.