The influence of lifestyle factors on survival following a diagnosis of colorectal cancer (CRC) is not well established. We examined associations between lifestyle factors measured before diagnosis and CRC survival. The Melbourne Collaborative Cohort Study collected data on alcohol intake, cigarette smoking and physical activity, and body measurements at baseline (1990–1994) and wave 2 (2003–2007). We included participants diagnosed to 31 August 2015 with incident stages I–III CRC within 10‐years post exposure assessment. Information on tumor characteristics and vital status was obtained. Tumor DNA was tested for microsatellite instability (MSI) and somatic mutations in oncogenes BRAF (V600E) and KRAS. We estimated hazard ratios (HRs) for associations between lifestyle factors and overall and CRC‐specific mortality using Cox regression. Of 724 eligible CRC cases, 339 died (170 from CRC) during follow‐up (average 9.0 years). Exercise (non‐occupational/leisure‐time) was associated with higher CRC‐specific survival for stage II (HR = 0.25, 95% CI: 0.10–0.60) but not stages I/III disease (p for interaction = 0.01), and possibly for colon and KRAS wild‐type tumors. Waist circumference was inversely associated with CRC‐specific survival (HR = 1.25 per 10 cm increment, 95% CI: 1.08–1.44), independent of stage, anatomic site and tumor molecular status. Cigarette smoking was associated with lower overall survival, with suggestive evidence of worse survival for BRAF mutated CRC, but not with CRC‐specific survival. Alcohol intake was not associated with survival. Survival did not differ by MSI status. We have identified pre‐diagnostic predictors of survival following CRC that may have clinical and public health relevance. What's new? While survival following colorectal cancer (CRC) is improving, it remains unclear whether increasing survival is linked to specific health or lifestyle factors that were present before or after CRC diagnosis. In our study, based on data from the Melbourne Collaborative Cohort Study (MCCS), exercise prior to diagnosis was associated with higher CRC‐specific survival specifically for stage II patients. Waist circumference, by contrast, was associated with reduced CRC‐specific survival, while cigarette smoking was associated with lower overall survival. Cigarette smoking was also weakly linked to decreased survival in BRAF‐mutated CRC. No associations were detected between CRC survival and alcohol intake.