A number of questionnaires have been used to assess the health-related quality of life of patients with chronic obstructive pulmonary disease (COPD). The study compares the performance of the SF-36 and the Nottingham Health Profile (NHP) in a sample of 321 male patients with COPD. Score distributions, reliability estimates, and several item scaling tests, including Rasch analysis were compared. Finally, we assessed the relative ability of the two instruments considered in discriminating different levels of disease severity by comparing: (a) their correlations with dyspnea and %FEV1; the receiver operating characteristic (ROC) curves; and the F-statistics. The SF-36 scores were less skewed and more homogeneously distributed than NHP scores. Item scaling tests and reliability estimates showed a better performance of SF-36 items and scales. Nevertheless, results of Rasch analysis evidenced that both instruments have very similar scaling characteristics. Validity results did not show a consistent pattern of superiority for either of the instruments. For the physical domain, correlations of NHP and SF-36 scores with %FEV1 and dyspnea were very similar and substantial (tau > or = 0.30). F-statistics showed that SF-36 physical scale was more precise (86%) than the NHP counterpart in discriminating among levels of dyspnea and %FEV1 impairment. Nevertheless, the ROC curve and its associated area under the curve did not show a significant difference (p > 0.10). Overall results suggest that both instruments are similar in discriminating among different levels of respiratory impairment.