Abstract In 2005, the National Institutes of Health (NIH) proposed standard criteria for diagnosis, organ scoring and global assessment of chronic graft-versus-host disease (cGvHD) severity. We retrospectively reclassified cGvHD with NIH criteria in a monocentric cohort of 130 consecutive adult patients with hematological malignancies presenting cGvHD after receiving allo-hematopoietic stem cell transplant (HSCT) with a fludarabine-busulfan-antithymocyte globulin (ATG) conditioning regimen, among 313 consecutive HSCT recipients. We compared NIH and Seattle classifications to correlate severity and outcome. The follow up range was effectively 2-120 months. Forty-four percent developed Seattle-defined cGvHD (22% limited, 78% extensive forms). Using NIH criteria, there were 23%, 40% and 37% mild, moderate and severe forms, respectively, and 58%, 32% and 8% classic cGvHD, late acute GvHD and overlap syndrome. Five-year overall survival was 55% (49-61), and cumulative incidences of non-relapse mortality (NRM) and relapse/progression at 2 years were 19% (14-23) and 19% (14-24). NIH mild and moderate forms were associated with better survival compared to severe cGvHD (hazard ratio [HR] = 3.28, 95% confidence interval [CI]: 1.38-7.82, p = 0.007), due to higher NRM among patients with severe cGvHD (HR = 3.04, 95% CI: 1.05-8.78, p = 0.04) but comparable relapse risk (p = NS). In conclusion, the NIH classification appears to be more accurate in predicting outcome mostly by the reclassification of old-defined extensive forms into NIH-defined moderate or severe.