Abstract. Myosteatosis, the infiltration of fat in skeletal muscle, is associated with lower skeletal muscle density (SMD) as detected by computed tomography (CT). It increases with aging and obesity and is thought to play a role in the aetiology of insulin resistance and type II diabetes. The clinical significance of myosteatosis in cancer cachexia, however, remains to be determined. Along with demonstrable subcutaneous and visceral lipolysis, myosteatosis may also be a key component of the syndrome. We aimed to investigate the use of human urine as a non‑invasive way to screen for molecular biomarkers of myosteatosis/reduced SMD using SELDI‑TOF mass spectrometry. Pre‑operative CT scans of patients undergoing surgery for upper gastrointestinal or hepatopancreaticobiliary cancer were analysed at the level of the third lumbar vertebrae. Myosteatosis was inferred as the presence of reduced SMD, which was defined as Hounsfield units for skeletal muscle <39.5 (two standard deviations below a normal healthy cohort). Urine was analysed by mass spectrometry using CM10 and IMAC30 SELDI‑chips. Peaks observed in the CM10 and IMAC30 chip types, showed marked expressional differences between control and myosteatosis, were further investigated by mascot SELDI matrix matching. A total of 55 patients was recruited; 31 patients were found to be myosteatotic on CT scan. Application of the IMAC30‑derived model to the entire cohort showed a sensitivity of 97%, specificity of 71% and an overall correctness of 85%. Application of the CM10 chipset‑based model to the entire cohort, showed a 77% sensitivity, 67% specificity and 73% overall correctness. Analysis of the peaks of interest resulted in the identification of significant fragments of cathepsin C, argin, arylsulfatase A and glial fibrillary acidic protein. We identified several potential urinary molecular biomarkers associated with reduced SMD in cancer. Such markers are potentially useful in deriving a clinical screening test for myosteatosis.