The Beck Depression Inventory (BDI) has been widely used to document the prevalence of depressive symptomatology in samples of chronic pain patients and as an outcome measure in studies of the psychological management of chronic pain. Several BDI items have a somatic content (sleep disturbance, fatigue, etc). Since chronic pain may have similar somatic effects, the significance of the total BDI score in this population is unclear. Two hundred and forty mixed chronic pain patients completed the BDI at screening interview for a psychologically based pain management programme; of these, 207 (mean age 50.1 years; 63% female) were later admitted to the programme when the BDI was readministered along with measures of pain, anxiety, pain-related cognitions, and physical performance. The mean BDI score of the sample was 18.1, with 71.7% meeting criteria (scoring 13 or more) for at least mild depression. A principal-components analysis of the BDI yielded 3 meaningful factors labelled: sadness about health, self-reproach, and somatic disturbance. Among the most frequently endorsed items were those loading on the somatic factor. The pattern of relationships between individual factor scores and measures of pain, mood, cognition, and physical functioning indicated that the use of the total BDI score may give a misleading impression of the nature and degree of affective disturbance in this group of patients. The implications of these findings for our understanding of BDI scores obtained by chronic pain patients are discussed.