BACKGROUND: Observational studies have shown that physical activity levels are inversely, and sedentary behaviours are positively, associated with colorectal cancer risk; however, whether these relationships are consistent across anatomical subsites is uncertain. METHODS: We investigated the associations between colorectal cancer and physical activity (metabolic equivalents (METs)-hours per week), and indicators of sedentary behaviour (television watching time and time spent using computers) among 430 584 men and women enroled in the UK Biobank. Multivariable hazard ratios (HRs) and 95% confidence intervals (CI) were estimated using Cox proportional hazards models. RESULTS: After a median follow-up time of 5.6 years, 2391 incident colorectal cancer cases were recorded. High (⩾60-MET-hours per week) vs low (<10-MET-hours per week) total physical activity was associated with a lower colon cancer risk (HR=0.84, 95% CI: 0.72-0.98; p-trend=0.04), with comparable relationships observed for proximal and distal colon tumours, but no association for rectal cancer. Higher levels of television watching time were associated with greater colon cancer risk (HR for ⩾5 h per day vs ⩽1 h per day=1.32, 95% CI: 1.04-1.68; p-trend=0.007). Time spent using computers was not associated with colorectal cancer risk. CONCLUSIONS: Higher levels of physical activity were associated with lower colon cancer risk, with no heterogeneity by colonic subsite. Sedentary behaviour (television watching) was associated with elevated colon cancer risk.British Journal of Cancer advance online publication, 8 March 2018; doi:10.1038/bjc.2017.496 www.bjcancer.com.