We investigated the effects of extinction and reinstatement on attentional bias to fear-conditioned signals in healthy individuals using an emotional modification of a spatial cueing paradigm. Spatial cues were emotionally modulated using differential conditioning. The CS+ was sometimes followed by an aversive electrocutaneous stimulus (UCS), whereas the CS- was never followed by the UCS. During a subsequent extinction phase no UCS was presented anymore. The reinstatement phase started with one or four unpredicted UCS-only trials for half of the participants (reinstatement group). For the other half there were no additional UCS presentations (control group). We found that attention was biased to threat signals during acquisition. This biased attention largely disappeared during extinction. During the reinstatement phase attentional bias to threat signals re-emerged in the reinstatement group, but not in the control group.