Attentional biases to threat are thought to play a central role in the development and maintenance of anxiety disorders. Various measures have been developed to index these biases; unfortunately, the psychometric properties of response time measures of attentional bias have not been thoroughly evaluated, and the reliability of those that have been evaluated is poor. The current study assessed the reliability and convergent validity of dot probe and emotional cueing bias scores in high-anxious and low-anxious participants in the context of a state anxiety manipulation to increase the likelihood of producing large, potentially reliable attentional biases. One hundred sixty participants completed an anxious or calm mood induction followed by dot probe and emotional cueing tasks. Reliability estimates for bias scores in the full sample were low, ranging from 0 to .44; convergent validity estimates were also poor. The anxiety induction did not substantially improve the reliability or validity of the measures. These results underscore the importance of developing new, more reliable attentional bias measures for future research.