Background Pirfenidone slows the progression of idiopathic pulmonary fibrosis. We investigated the efficacy and safety of pirfenidone by dose and disease severity in real-world patients with idiopathic pulmonary fibrosis.Methods This multi-centre retrospective cohort study investigated 338 patients treated with pirfenidone between July 2012 and March 2018. Demographics, pulmonary function, mortality, and pirfenidone-related adverse events were recorded. Efficacy was analysed according to pirfenidone dose and disease severity using linear mixed-effects models to assess the annual decline rate of forced vital capacity (FVC) and diffusing capacity (DL CO ) of the lungs for carbon monoxide.Results The mean %FVC predicted and %DL COpredicted were 72.6±13.1% and 61.4±17.9%, respectively. Pirfenidone treatment lasted 16.1±9.0 months. In the standard-dose (1800 mg/day) group, the mean %FVC predicted was -6.56% (95% CI -9.26, -3.87) per year pre-pirfenidone treatment, but -4.43% (95% CI -5.87, -3.00) per year post-treatment. In the lower-dose group, the mean %FVC predicted was -4.96% (95% CI -6.82, -3.09) per year pre-pirfenidone treatment, but -1.79% (95% CI -2.75, -0.83) per year post-treatment. The FVC decline rate was significantly reduced regardless of GAP stage. However, the DL CO decline rate was significantly reduced in the GAP stage II–III group, but not stage I group. Adverse events and mortality were similar across dose groups, but were more frequent in the GAP stage II–III group than in the stage I group.Conclusions The effect of pirfenidone on reducing disease progression persisted even with a consistent lower dose of pirfenidone.Trial registration Retrospectively registered