The selection criteria of hepatectomy for patients with multiple hepatocellular carcinomas (HCCs) remain controversial. A scoring system based on preoperative data and independent predictors of overall survival (OS) was developed in a primary cohort of 510 patients who underwent hepatectomy for multiple HCCs from 1998 to 2006, and validated in 177 patients who were operated from 2006 to 2009 at the Eastern Hepatobiliary Surgery Hospital. In the NDR scoring system, tumor number (N) > 3, total tumor diameter (D) > 8 cm, and a ratio of largest/smallest diameter (R) > 6 were independent predictors of OS. Its predictive accuracy as determined by the area under the curve (AUC, 0.718) was larger than the four conventional staging systems (0.524-0.662). It stratified postoperative OS into five levels (0-4 score). The 5-year OS rate of patients with a NDR score 0-2 was 46.5% versus 13.9% in those > 2 (P < 0.001). Patients with a score 0-2 therefore were recommended for hepatectomy. The feasibility of this NDR score 0-2 was compared with the previously reported criteria. If the two more stringent inclusion criteria were adopted, 45.5-75.7% of patients with a NDR score 0-2 would be excluded, but their 5-year OS rates were comparable to those within the criteria (44.7% vs. 52.1%, P = 0.083; 46.6% vs. 46.3%, P = 0.674). If the less stringent criteria were used, an additional 25.9% of patients received hepatectomy, but their 5-year OS rate was 13.9%. The NDR scoring system was more accurate in selecting patients with multiple HCCs for hepatectomy.