Epidemiological studies have suggested positive associations between diabetes and colorectal cancer. However, the findings were inconsistent, especially regarding specific sex or cancer subsite. To conduct a systematic review with meta-analysis to examine the association of diabetes mellitus with risk of colorectal cancer and to investigate whether the association is dependent on sex, cancer subsite, race/ethnicity, or duration of diabetes. Relevant studies were identified by searching Pubmed to May 12, 2011. The reference lists of identified articles were also reviewed. Two authors independently selected the studies and extracted relevant data. The study quality was assessed. Thirty-nine studies (28 cohort studies and 11 case-control studies) were included in the final analysis. Compared with non-diabetic subjects, the pooled risk estimate of colorectal cancer for diabetic subjects was 1.29 (95% CI = 1.23-1.35). The subgroup analyses revealed that the risk was significantly increased for both genders, for every cancer subsite, and for both short duration and long duration of diabetes. Exposure to diabetes increased the risk of colorectal cancer for White and Asian, whereas there was no evidence of such an association either among Black or native Hawaiians. Our results confirmed that subjects with diabetes are at modestly increased risk of developing colorectal cancer.