A considerable volume of research has demonstrated an anxiety-linked attentional bias characterized by selective processing of threat stimuli. The last decade has seen growing interest in identifying the precise attentional mechanisms which underlie such selective processing to advance both theoretical and etiological models of anxiety. This research has particularly focused on the roles of spatial engagement and disengagement of attention. The relative contribution of these attentional components to selective processing of threat in anxious individuals remains unclear however. Moreover, we argue here that many of the tasks employed to examine these mechanisms, may not be capable of indexing the attentional processes that they claim to measure. In this article, we provide a methodological review, critically evaluating the adequacy of previous tasks employed to measure biased attentional engagement and disengagement. Based on a number of concerns raised about the ability of such tasks to differentiate these component attentional processes, we detail three task criteria that we believe are essential to be confident that a task will accurately index biased attentional engagement with, and disengagement from threat in anxious participants.