We used the National Institutes of Health (NIH) criteria for the diagnosis, classification and scoring of chronic GVHD (cGVHD) to reevaluate patients with cGVHD originally diagnosed using classic criteria. We retrieved data from 236 patients diagnosed with cGVHD on the basis of classic criteria. Excluding 20 'liver-alone' patients, we re-categorized 216 patients in keeping with the NIH criteria. Twenty patients were reclassified as having acute GVHD and 196 patients as having cGVHD (170 'classic chronic' (Cl-Ch) and 26 'overlap chronic' (Ov-Ch)). The 5-year GVHD-specific survival (GSS) was significantly different between the two cGVHD subtypes, specifically 87.3% for Cl-Ch vs 70.2% for Ov-Ch (P=0.006). The NIH severity criteria were effective in expecting 5-year GSS rates at both the onset (93.5, 81.3 and 79.7% (P=0.047)) and peak intensity of the disease (100, 89.7 and 78.7% (P=0.004) for the mild, moderate and severe grade, respectively). Multivariate analysis showed that NIH severity criteria were independently significant prognostic factors for GSS (mild vs moderate, HR 4.35, P=0.036; mild vs severe, HR 5.25, P=0.020). Our results support the role of the NIH criteria in classifying cGVHD and in assessing the severity of the disease to predict patient prognosis of cGVHD.