Abstract: Objective To study the attention biases to pain-related information in chronic pain patients. Methods Attention biases in 20 chronic pain patients and 20 healthy controls to pain-related words were recorded with E-prime 2.0 program, Stroop task. Results The reaction time (RT) to pain-related words and threaten words was longer than to neutral words in chronic pain patients(786.02±18.15 ms vs 774.73 ±17.34 ms, 792.35±17.72 ms vs 774.73±17.34 ms, P< 0.05). The RT to threaten-related words was longer than to neutral words in healthy controls (779.75±20.99 ms vs 764.74±20.53 ms, P< 0.05) whereas no significant difference was found in pain-related words and neutral words. Variance analysis showed a dominant effect of stimulation types (F=18.86, P< 0.01)and an interactive effect between group types and stimulation types (F= 4.48，P< 0.05). Pain-related words were not related with pain-lasting time but with anxiety in chronic pain patients with attention biases(r=0.471, P< 0.05). Conclusion Chronic pain patients suffer attention biases to pain-related information, which is significantly related to their anxiety.