Objectives: Hepatic grading systems for categorizing severity in chronic graft-versus-host disease (cGvHD) were determined arbitrarily, leading us to initiate the present study to provide objective evidence for the determination of optimal cutoff values and devise a hepatic grading system to predict prognosis.Methods: Of 842 patients who received allogeneic hematopoietic stem transplant (HCT), 336 patients diagnosed with cGvHD were evaluated for overall survival (OS) and non-relapse mortality (NRM) after cGVHD development. Multiple statistical parameters were evaluated to define optimal cutoff values of liver profile, including negative predictive value (NPV), positive predictive value (PPV), accuracy, and p-values as measures of risk stratification power. Results: We found that alkaline phosphatase (ALP) ≥ 146 IU/L (NPV: 83.4%; PPV: 32.8%; accuracy: 52.7%) and bilirubin ≥ 14 µmol/L (NPV: 81.8%; PPV: 39.4%; accuracy 68.1%) significantly correlated with OS. We developed a refined hepatic cGvHD grading score (RHS), stratifying patients into a low-RHS group with RHS score 0, OS at 3 years (n = 162) to 80.5%, compared to high-RHS group with score 1-2 (n = 172) 62.7%. Regarding NRM, score 0 segregated NRM at 3 years to 11.9%, compared with score 1-2 19.6%, P = .1.Conclusions: Refined hepatic score is promising for stratifying patients with cGVHD and liver involvement according to long-term outcomes.