Background/objectives Traditional malnutrition screening instruments, including the Malnutrition Universal Screening Tool (MUST), strongly rely on low body mass index (BMI) and weight loss. In overweight/obese patients, this may result in underdetection of malnutrition risk. Alternative instruments, like the Patient-Generated Subjective Global Assessment Short Form (PG-SGA SF), include characteristics and risk factors irrespective of BMI. Therefore, we aimed to compare performance of MUST and PG-SGA SF in malnutrition risk evaluation in overweight/obese hospitalized patients. Subjects/methods We assessed malnutrition risk using MUST (>= 1 = increased risk) and PG-SGA SF (>= 4 = increased risk) in adult patients at hospital admission in a university hospital. We compared results for patients with BMI < 25 kg/m(2) vs. BMI >= 25 kg/m(2). Results Of 430 patients analyzed (58 +/- 16 years, 53% male, BMI 26.9 +/- 5.5 kg/m(2)), 35% were overweight and 25% obese. Malnutrition risk was present in 16% according to MUST and 42% according to PG-SGA SF. In patients with BMI < 25 kg/m(2), MUST identified 31% as at risk vs. 52% by PG-SGA SF. In patients with BMI >= 25 kg/m(2), MUST identified 5% as at risk vs. 36% by PG-SGA SF. Agreement between MUST and PG-SGA SF was low (k = 0.143). Of the overweight/obese patients at risk according to PG-SGA SF, 83/92 (90%) were categorized as low risk by MUST. Conclusions More than one-third of overweight/obese patients is at risk for malnutrition at hospital admission according to PG-SGA SF. Most of them are not identified by MUST. Awareness of BMI-dependency of malnutrition screening instruments and potential underestimation of malnutrition risk in overweight/obese patients by using these instruments is warranted.