Background: The optimal strategy for CRC post-treatment surveillance is unknown. Existing guidelines are based on limited evidence and not risk-based. The purpose of this study is to examine surveillance patterns and intensity in routine practice and the impact on recurrence detection. Methods: Primary records of a random sample of 10,885 Stage I-III CRC patients from Commission on Cancer accredited hospitals (2006-2007) were individually abstracted and detailed results of surveillance testing (advanced imaging, endoscopy and CEA) were reviewed. Data was merged with records in the National Cancer Database (NCDB). Incidence rate ratios (IRR) of tests were determined using multivariable negative binomial regression with an offset variable. SI quartiles were identified based on recurrence risk-independent factors and rates of recurrence detection assessed using Cox regression. Results: Testing rates per person-year were 0.69 (imaging), 0.22 (endoscopy), and 0.77 (CEA). No testing was performed for 20%, 17% and 13% of stage I, II, and III patients. Advanced stage (III vs I) was associated with imaging (IRR = 1.87), endoscopy (IRR = .8) and CEA (IRR = 1.68). After adjusting for cancer-related factors, SI of imaging (p = .007, Q4 vs Q1) but not endoscopy or CEA (all p > .1) was associated with detection of distant recurrence. The association with imaging SI was limited to stage III patients in the highest SI quartile only and insignificant for all other groups. Conclusions: Variation exists in the intensity of surveillance following curative treatment of CRC in routine practice. While there is significant variation associated with underlying recurrence risk (e.g. stage), after accounting for cancer-related factors, surveillance intensity was associated with rate of recurrence detection only among the highest risk cohort. These data inform the development of optimized, risk-stratified approaches to CRC surveillance.