Cognitive models of anxiety propose that selective attention to negative information plays a causal role in heightened anxiety vulnerability and dysfunction. However, there has been theoretical disagreement concerning whether anxiety-linked attentional biases reflect enhanced attentional engagement with, or impaired attentional disengagement from, negative information. We contend that previous methodologies have not been optimal in terms of their capacity to differentiate both types of bias. The present study introduces a refined methodology, in which the conventional dot-probe task is modified in a novel manner to enable the independent assessment of these components of attention. The findings demonstrate that facilitated attentional engagement and impaired attentional disengagement are both characteristic of elevated levels of anxiety vulnerability. Moreover, these prove to be unrelated facets of attentional selectivity that independently contribute to variation in anxiety vulnerability. We discuss the possibility that these two types of attentional bias may represent independent pathways to anxiety vulnerability.