Our previous work suggests that somatic threat feature detectors indexed by a pain-evoked midlatency negative scalp potential play an important role in the attentional bias toward pain. In these studies the somatic threat feature detectors facilitated the shift in attention to a somatic threat when attention was focused on another stimulus modality but not when it was focused on another spatial location. This experiment used the Posner cuing paradigm to investigate possible explanations for this discrepancy. The results demonstrate that the different somatic threat effects observed in previous modal and spatial cuing studies are not due to attentional set. Rather, this discrepancy may be related to differences in contingent attention capture. This study also verifies earlier findings suggesting that some of the pain-related P2 component generators are involved in stimulus-driven shifts in spatial attention toward task-relevant stimuli.