Purpose <p>Sarcopenia is characterized by loss of skeletal muscle mass, strength, and physical ability and is a well-established predictor of postoperative complications. However, the optimal method for preoperative assessment remains controversial. This study evaluated preoperative skeletal muscle density using computed tomography as a surrogate marker for sarcopenia to predict postoperative complications in patients undergoing highly invasive surgery for upper abdominal cancer.</p> Methods <p>We examined 46 cases (esophagectomy, n = 33; pancreaticoduodenectomy, n = 13). We examined the relationship between clinical parameters and indicators of physical performance (peak load, six-minute walking test) and analyzed the impact of preoperative skeletal muscle density on postoperative complications.</p> Results <p>Muscle density was significantly associated with physical performance in both parameters (peak load, <Emphasis Type="BoldItalic">p</Emphasis> = 0.0007; six-minute walking test, <Emphasis Type="BoldItalic">p</Emphasis> = 0.0001). The optimal cutoff value of muscle density for physical performance was 31.4 HU, with a sensitivity of 83.3% and specificity of 80% (area under curve = 0.833). In the multivariate analysis, the presence of low muscle density was a significant risk factor for postoperative complications (<Emphasis Type="BoldItalic">p</Emphasis> = 0.0225).</p> Conclusion <p>The preoperative skeletal muscle density may serve as a simple and objective surrogate marker for sarcopenia. It might be a useful predictor of postoperative complications after highly invasive surgery for upper abdominal cancer.</p>