Background The best method to interpret the chest roentgenogram and its sensitivity to detect effect of treatment for sarcoidosis remains unclear. In a double blind, randomized trial of infliximab for chronic pulmonary sarcoidosis, changes in serial chest roentgenograms were examined by radiologists, blinded to order or treatment. Methods Chest roentgenograms were at weeks 0, 6, and 24 weeks of therapy with either placebo or 3 or 5 mg/kilogram infliximab. Films were reviewed in random order by two independent radiologists, unaware of treatment. The films were compared using two methods: the prespecified objective assessment - a scoring system previously proposed by Muers, and the post hoc assessment - a 5 point Likert scale global assessment between two films. Results Of 138 patients enrolled in the study, chest roentgenograms for all studies were available on 130 patients. There was only fair agreement between the two radiologists in the original stage of the patient's chest roentgenogram (weighted kappa=0.43, 95% CI = 0.32-0.54). For the Likert scale of global assessment of change, there was good agreement between the two readers (weighted kappa=0.61, 95% CI = 0.51-0.71). There was good correlation between the two readers for the various components of the Muers' score, especially the reticulonodular (R) score (R = 0.578, p < 0.05). The initial R score was positively correlated with improvement in FVC with infliximab therapy (R = 0.239, p < 0.05). Conclusion Global assessment and Muers' scoring system were associated with good agreement between two expert readers. Improvement in both scores correlated with improvement in FVC.