Background & aims: Sarcopenia is associated with an increased risk of complications to treatment and lower survival rates in patients with cancer, but there is a lack of agreement on cut-off values and assessment methods. We aimed to investigate the prevalence of sarcopenia assessed by dual-energy x-ray absorptiometry (DXA) and computed tomography (CT) as well as the agreement between the methods for identification of sarcopenia. Methods: This cross-sectional study pooled data from two studies including patients scheduled for surgery for gastrointestinal tumors. We assessed sarcopenia using two different cut-off values derived from healthy young adults for DXA and two for CT. Additionally, we used one of the most widely applied cut-off values for CT assessed sarcopenia derived from obese cancer patients. The agreement between DXA and CT was evaluated using Cohen's kappa. The mean difference and range of agreement between DXA and CT for estimating total and appendicular lean soft tissue were assessed using Bland–Altman plots. Results: In total, 131 patients were included. With DXA the prevalence of sarcopenia was 11.5% and 19.1%. Using CT, the prevalence of sarcopenia was 3.8% and 26.7% using cut-off values from healthy young adults and 64.1% using the widely applied cut-off value. The agreement between DXA and CT in identifying sarcopenia was poor, with Cohen's kappa values ranging from 0.05 to 0.39. The mean difference for estimated total lean soft tissue was 1.4 kg, with 95% limits of agreement from −8.6 to 11.5 kg. For appendicular lean soft tissue, the ratio between DXA and CT was 1.15, with 95% limits of agreement from 0.92 to 1.44. Conclusions: The prevalence of sarcopenia defined using DXA and CT varied substantially, and the agreement between the two modalities is poor.