Preliminary evidence from a study using a modified Stroop paradigm suggests that individuals with chronic pain selectively attend to pain-related information. The current study was conducted in an attempt to replicate and extend this finding. Nineteen patients with chronic pain stemming from musculoskeletal injury and 22 healthy control subjects participated. All participants completed a computerised task designed to evaluate attentional allocation to cues thematically related to pain and injury via measurement of detection latencies for dot-probes that followed their presentation. Results indicated that patients did not differ from control subjects in their pattern of responses to dot-probes that were presented following either the pain- or injury-related cues. This pattern of results continued to hold true after including level of depression as a covariate in the analysis. However, when patients were divided on the basis of scores on the Anxiety Sensitivity Index (Peterson, R.A. and Reiss, S., Anxiety Sensitivity Index Manual, 2nd edn., International Diagnostic Systems, Worthington, OH, 1992), a measure related to fear of pain (Asmundson, G.J.G. and Norton, G.R., Behav. Res. Ther., 34 (1996) 545-554), those with low anxiety sensitivity shifted attention away from stimuli related to pain whereas those with high anxiety sensitivity responded similarly to dot-probes regardless of the parameters of presentation. These results suggest that the operation of the information processing system in patients with chronic pain may be dependent on a patient's trait predisposition to fear pain. Theoretical and ecological implications are discussed.