This prospective cohort study was conducted in eighteen Canadian hospitals with the aim of examining factors associated with nutritional decline in medical and surgical patients. Nutritional decline was defined based on subjective global assessment (SGA) performed at admission and discharge. Data were collected on demographics, medical information, food intake and patients’ satisfaction with nutrition care and meals during hospitalisation; 424 long-stay (≥7 d) patients were included; 38 % of them had surgery; 51 % were malnourished at admission (SGA B or C); 37 % had in-hospital changes in SGA; 19·6 % deteriorated (14·6 % from SGA A to B/C and 5 % from SGA B to C); 17·4 % improved (10·6 % from SGA B to A, 6·8 % from SGA C to B/A); and 63·0 % patients were stable (34·4 % were SGA A, 21·3 % SGA B, 7·3 % SGA C). One SGA C patient had weight loss ≥5 %, likely due to fluid loss and was designated as stable. A subset of 364 patients with admission SGA A and B was included in the multiple logistic regression models to determine factors associated with nutritional decline. After controlling for SGA at admission and the presence of a surgical procedure, lower admission BMI, cancer, two or more diagnostic categories, new in-hospital infection, reduced food intake, dissatisfaction with food quality and illness affecting food intake were factors significantly associated with nutritional decline in medical patients. For surgical patients, only male sex was associated with nutritional decline. Factors associated with nutritional decline are different in medical and surgical patients. Identifying these factors may assist nutritional care.