We argue that the conflicting results reported in previous studies examining the factor structure of the McGill Pain Questionnaire Pain Rating Index (PRI) can be explained by differences in the patient samples and statistical analyses used across studies. In an effort to clarify the factor structure of the PRI, 3 different factor models were compared using confirmatory factor analysis in 2 samples of low-back pain patients (N = 1372) and in a third sample of patients suffering from other chronic pain problems (N = 423). A 4-factor model, similar to those obtained in previous studies where multiple criteria were used to determine the number of factors extracted, best explained covariation among PRI subclasses. However, relatively high interfactor correlations (approximately two-thirds of the variance explained by the best fitting factor structure was common variance) cast doubt on the discriminant validity of PRI subscales; examination of relationships between the PRI and MMPI subscales also failed to provide evidence of the discriminant validity or clinical utility of PRI subscales. Reducing the information from the 10 PRI sensory subclasses to a single subscale score may seriously limit the usefulness of the PRI. Alternate methods of using PRI data are suggested. © 1992.