Background: External beam radiation therapy has been used as a palliative to radical treatment of hepatocellular carcinoma (HCC) depending on different tumor status, liver function and patient's general state of health. The existing models of HCC staging cannot perfectly predict the prognosis of radiotherapy. In this study, we aimed to set up a new staging system for radiotherapy-based treatment by incorporating bilirubin-albumin (ALBI) grade and tumor status for the prognostic classifications of HCC. Methods: This multicenter cohort study included 878 HCC patients who received radiotherapy-based treatment. A new staging system was established: stage I, solitary nodule without macrovascular invasion or 2-3 nodules with no more than 3.0 cm each other and PS 0-2 (Ia: ALBI-1 grade; Ib: ALBI-2 or 3 grade); stage II: 2-3 nodules with anyone more than 3.0 cm or ≥4 nodules and PS 0-2 (IIa: ALBI-1 grade; IIb: ALBI-2 grade); stage III: macrovascular invasion or regional lymph node metastasis or distant metastasis and PS 0-2 (IIIa: ALBI-1 grade; IIIb:ALBI-2 grade); stage IV: ALBI-3 grade without stage I patient or/and PS score 3-4. The new modified staging system and the existing staging systems, such as the BCLC, TNM, CNLC staging systems were used for prognostic analysis. All patients were separated into different stages and substages. The long-term overall survival outcomes and time-dependent receiver operating characteristic (ROC) were analyzed. Results: A training cohort of 595 patients underwent stereotactic body radiotherapy (SBRT) from 2011 to 2017 and an external validation cohort of 283 patients underwent intensity-modulated radiotherapy (IMRT) from 2000 to 2013 were included into establishing and validating the new staging system. In the training cohort, the median follow-up time was 55 months (range, 6–100 months), and the new staging system had a good discriminatory ability to separate patients into different stages with 4 notably different curves and substages with 7 notably different curves. BCLC staging could not differentiate stage 0 to A, and stage C to D in these selected patients. TNM staging could not completely distinguish stage IIIb to IV, but also stage Ia to Ib. CNLC staging could not differentiate among stage IIIa, IIIb, and IV. In the external validation, the median follow-up time was 95 months (range, 9–120 months), and the new staging system also had a good discriminatory ability to separate patients into different stages with 4 notably different curves and substages with 7 notably different curves. The new staging system had a better area under curve of time-dependent ROC than BCLC, TNM and CNLC staging in both SBRT and IMRT cohorts. Conclusions: The new modified (Su’s) staging system could provide a good discriminatory ability to separate patients into different stages and substages after radiotherapy treatment. It may be used to supplement the other HCC staging systems.