Background. Physical activity (PA) reduces incidence of colorectal cancer (CRC). Its influence on cancer-specific (CSS) and overall survival (OS) is controversial. Methods. We performed a literature-based meta-analysis (MA) of observational studies, using keywords "colorectal cancer, physical activity, and survival" in PubMed and EMBASE. No dedicated MA was found in the Cochrane Library. References were cross-checked. Pre- and postdiagnosis PA levels were assessed by MET. Usually, "high" PA was higher than 17 MET hour/week. Hazard ratios (HRs) for OS and CSS were calculated, with their 95% confidence interval. We used more conservative adjusted HRs, since variables of adjustment were similar between studies. When higher PA was associated with improved survival, HRs for detrimental events were set to <1. We used EasyMA software and fixed effect model whenever possible. Results. Seven studies (8056 participants) were included, representing 3762 men and 4256 women, 5210 colon and 1745 rectum cancers. Mean age was 67 years. HR CSS for postdiagnosis PA (higher PA versus lower) was 0.61 (0.44-0.86). The corresponding HR OS was 0.62 (0.54-0.71). HR CSS for prediagnosis PA was 0.75 (0.62-0.91). The corresponding HR OS was 0.74 (0.62-0.89). Conclusion. Higher PA predicted a better CSS. Sustained PA should be advised for CRC. OS also improved (reduced cardiovascular risk). © 2013 Gaetan Des Guetz et al.