Various versions of the probe detection task have been developed to assess attentional biases in anxiety and there is debate about their relative merits in terms of reliability and sensitivity to such biases. The present study used a pictorial version of the probe detection task to examine attentional biases for emotional facial expressions. The main aims were (1) to see if our previous finding of greater vigilance for threatening faces in high than low trait anxiety could be replicated [Bradley, B. P., Mogg, K., Falla, S. J., & Hamilton, L. R. (1998). Attentional bias for threatening facial expressions in anxiety: manipulation of stimulus duration. Cognition and Emotion, in press] and (2) to examine whether the same pattern of results and a similar effect size, would be obtained using a 'probe position' task (i.e. where is the probe?), rather than the 'probe classification' task (i.e. what is the type of the probe?) used by Bradley et al. (1998). The probe position task produced similar results to those obtained from the probe classification task, so providing further evidence of an anxiety-related attentional bias for threatening faces. Results also indicated that, for non-clinical participants, the probe position task yielded faster overall RTs, fewer errors and a similar effect size, compared with the probe classification task. Implications for the assessment of attentional biases in non-clinical and clinical samples are discussed.