Purpose To assess whether preoperative radiologically defined lean muscle measures are associated with adverse clinical outcomes in patients undergoing elective surgery for colorectal cancer. Methods This retrospective UK-based multicentre data collection study identified patients having had colorectal cancer resection with curative intent between January 2013 to December 2016. Preoperative computed-tomography (CT) scans were used to measure psoas muscle characteristics. Clinical records provided postoperative morbidity and mortality data. Results This study included 1122 patients. The cohort was separated into a combined group (patients with both sarcopenia and myosteatosis) and others group (either sarcopenia or myosteatosis, or neither). For the combined group, anastomotic leak was predicted on univariate (OR 4.1, 95% CI 1.43–11.79; p  = 0.009) and multivariate analysis (OR 4.37, 95% CI 1.41–13.53; p  = 0.01). Also for the combined group, mortality (up to 5 years postoperatively) was predicted on univariate (HR 2.41, 95% CI 1.64–3.52; p  < 0.001) and multivariate analysis (HR 1.93, 95% CI 1.28–2.89; p  = 0.002). A strong correlation exists between freehand-drawn region of interest-derived psoas density measurement and using the ellipse tool ( R 2  = 81%; p  < 0.001). Conclusion Measures of lean muscle quality and quantity, which predict important clinical outcomes, can be quickly and easily taken from routine preoperative imaging in patients being considered for colorectal cancer surgery. As poor muscle mass and quality are again shown to predict poorer clinical outcomes, these should be proactively targeted within prehabilitation, perioperative and rehabilitation phases to minimise negative impact of these pathological states.