Objective: to analyse the differences in malnutrition assessment between the Global Leadership Initiative on Malnutrition (GLIM) criteria and the Patient-Generated Subjective Global Assessment (PG-SGA) among patients with hepatobiliary and pancreatic malignancies. Method: this study was a cross-sectional study and included 126 hospitalised patients who underwent surgery for hepatobiliary and pancreatic malignancies between November 1, 2019 and August 1, 2020. The patients' clinical data were collected, and malnutrition assessments were completed using the different nutritional assessment tools. The consistency of both tools was analysed using Cohen's kappa coefficient. Results: the prevalence of malnutrition showed a difference in diagnosis results between the GLIM criteria (36.51 %) and the PG-SGA (55.56 %). The two methods had moderate consistency (kappa = 0.590, p < 0.01). The sensitivity of a malnutrition diagnosis using a combination of GLIM and PG-SGA was 65.7 % (53.3 % and 76.4 %, respectively), and specificity was 100 % (92 % and 100 %, respectively). When malnutrition was evaluated using only PG-SGA, sensitivity was 88.9 % (95 % confidence interval (CI) 63.9 % to 98.1 %), whereas when only the GLIM score was used for malnutrition evaluation, sensitivity was 98.2 % (95 % CI, 92.8 % to 99.7 %). In addition, the PG-SGA score and the GLIM score had significant correlations. Conclusion: GLIM performed better than PG-SGA in the correlation analysis of nutritional indicators. GLIM is more suitable for patients with hepatobiliary and pancreatic malignancies than PG-SGA.