Little is known about the value of the nutritional risk screening 2002 (NRS2002) scale in nasopharyngeal carcinoma (NPC). We conducted a large‐scale study to address this issue. We employed a big‐data intelligence database platform at our center and identified 3232 eligible patients treated between 2009 and 2013. Of the 3232 (12.9% of 24 986) eligible patients, 469 (14.5%), 13 (0.4%), 953 (29.5%), 1762 (54.5%) and 35 (1.1%) had NRS2002 scores of 1, 2, 3, 4 and 5, respectively. Survival outcomes were comparable between patients with NRS2002 <3 and ≥3 (original scale). However, patients with NRS2002 ≤3 vs >3 (regrouping scale) had significantly different 5‐year disease‐free survival (DFS; 82.7% vs 75.0%, P < .001), overall survival (OS; 88.8% vs 84.1%, P = .001), distant metastasis‐free survival (DMFS; 90.2% vs 85.9%, P = .001) and locoregional relapse‐free survival (LRRFS; 91.6% vs 87.2%, P = .001). Therefore, we proposed a revised NRS2002 scale, and found that it provides a better risk stratification than the original or regrouping scales for predicting DFS (area under the curve [AUC] = 0.530 vs 0.554 vs 0.577; P < .05), OS (AUC = 0.534 vs 0.563 vs 0.582; P < .05), DMFS (AUC = 0.531 vs 0.567 vs 0.590; P < .05) and LRRFS (AUC = 0.529 vs 0.542 vs 0.564; P < .05 except scale A vs B). Our proposed NRS2002 scale represents a simple, clinically useful tool for nutritional risk screening in NPC.