In the present study, we aimed to validate the Malnutrition Universal Screening Tool (MUST) for routine nutritional screening in the radiation oncology setting, thus enabling timely and adequate referrals of patients at risk for individualised or advanced intervention. Towards this objective, we conducted a prospective cross-sectional study in 450 non-selected cancer patients (18-95 years) referred for radiotherapy. The following were the nutritional parameters: BMI (categorised by WHO's age/sex criteria), weight loss >5 % in the previous 3-6 months, Patient-Generated Subjective Global Assessment (PG-SGA - validated/specific for oncology) and nutritional risk by MUST. Sensitivity, specificity, predictive values and concordance were calculated to validate MUST v. PG-SGA and compare single parameters v. PG-SGA/MUST. BMI v. PG-SGA showed a negligible capacity to detect undernutrition: 0.27 sensitivity, 0.23 specificity, 0.35 positive predictive value and 0.31 negative predictive value. Conversely, percentage weight loss v. PG-SGA was highly effective: 0.76 sensitivity, 0.85 specificity, 0.79 positive predictive value and 0.85 negative predictive value. MUST v. PG-SGA successfully detected patients at risk: 0.80 sensitivity, 0.89 specificity, 0.87 positive predictive value and 1.0 negative predictive value; percentage weight loss v. MUST proved able to identify patients likely to be at risk: 0.85 sensitivity, 0.91 specificity, 0.90 positive predictive value and 1.0 negative predictive value. This is the first study in the radiation oncology setting to validate MUST: a simple and quick method applicable by any health professional, with a high validity for early screening, ideally to antedate a comprehensive nutritional assessment and guide for intervention. In this study, percentage weight loss in the previous 3-6 months does seem valid to predict nutritional risk, and may be the minimum in a busy routine.