To investigate attentional bias to threatening information, the authors propose a new version of the spatial cueing paradigm in which the focus is on perceptual accuracy instead of response speed. In two experiments, healthy volunteers made unspeeded discriminations between three visual targets presented left or right. Each target was preceded by a visual cue (colored rectangle) at either the same (valid) or opposite (invalid) location. By means of differential classical conditioning with aversive white noise, a threat cue and a control cue were created. Analyses of error rates showed that cueing effects (lower proportion of errors in valid trials relative to invalid trials) were more pronounced in threat trials than in neutral trials. This threat-related bias was particularly because of threat cues reducing accuracy in invalid trials, indicating difficulty disengaging attention from threatening information. Engagement of attention was not affected by threat, as threat cues did not facilitate the processing of targets in valid trials. The findings are discussed in light of the strengths and limitations of spatial cueing tasks.