BACKGROUND AND OBJECTIVE: In Japan, the classification of disease severity of idiopathic pulmonary fibrosis (IPF) (J-system) has been used in making decisions on medical care subsidies. The present J-system consists of arterial partial pressure of oxygen (PaO2 ) and exercise desaturation in stages of I-IV. It provides a good prognostic classification in stages III and IV, but not in stages I and II. Therefore, we propose a revised system to improve discriminative ability in stages I and II. METHODS: We compared the revised J-system with the present J-system using Cox proportional hazards model to predict mortality rate. We also evaluated the recently proposed GAP (Gender, Age and Physiology) system in comparison to both J-systems. RESULTS: Two-hundred and fifteen IPF patients were studied retrospectively. A univariate model showed that the present and revised J-systems and a modified GAP system were all significant prognostic factors. The C-statistic for discriminating prognosis was higher in the revised J-system than the modified GAP system and the present J-system (0.677, 0.652 and 0.659, respectively). The C-statistics of these models produced from the 10 000 bootstrap samples were similar to those of the original models, suggesting good internal validation (0.665 (95% CI: 0.621-0.705), 0.645 (0.600-0.686) and 0.659 (0.616-0.700), respectively). Multivariate analysis revealed that the revised J-system (P = 0.0038) and the modified GAP system (P = 0.0029) were independent prognostic factors. CONCLUSION: The revised J-system can provide a better mortality prediction than the present one. Both the revised J-system and the modified GAP system are independent and valuable tools for prognostication and clinical management for IPF.