A major problem in the understanding and psychological treatment of chronic pain patients is the inadequacy of available assessment procedures. Currently, the most frequently used instrument is the Pain Rating Index (PRI) of the McGill Pain Questionnaire, designed to assess 3 components of pain (i.e., sensory, affective, and evaluative) postulated by the Gate Control Theory. The PRI has been used in many studies to make differential diagnoses, to describe different groups of pain patients, and to identify the factor composition of the instrument itself. To date, however, no study has appropriately tested the a priori structure of the PRI or cross-validated it. Confirmatory factor analytic procedures employed in the present study supported Melzack's postulated tri-component structure of the PRI in 2 diverse samples of pain patients from different hospitals. However, the 3 components were found to be highly intercorrelated. Subsequent analyses revealed that the 3 components of the PRI do not display adequate discriminant validity. The current use of 3 subscales to establish differential diagnoses or patterns of different pain syndromes may lead to inappropriate classification and treatment decisions. The major conclusion of these findings is that use of only the total score of the PRI is appropriate for pain assessment.