Purpose To assess the value of PET/CT for detecting local or distant recurrence in patients who undergo surgery for colorectal cancer (CRC) and to compare the accuracy of PET/CT to that of conventional imaging studies (CIS). Methods Tumor surveillance PET/CT scans done between March 2005 and December 2009 of disease-free patients after surgery with or without adjuvant chemotherapy for CRC were retrospectively studied. CIS (serial enhanced CT from lung base to pelvis and plain chest radiograph) were performed within 1 month of PET/CT. We excluded patients with distant metastasis on initial staging, a known recurrent tumor, and a lack of follow-up imaging. The final diagnosis was based on at least 6 months of follow-up with colonoscopy, biopsy, and serial imaging studies in combination with carcinoembryonic antigen levels. Results A total of 262 PET/CT scans of 245 patients were included. Local and distant recurrences were detected in 27 cases (10.3%). On case-based analysis, the overall sensitivity, specificity, and accuracy were 100, 97.0, and 97.3% for PET/CT and 85.1, 97.0, and 95.8% for CIS, respectively. On lesion-based analysis, PET/CT detected more lesions compared to CIS in local recurrence and lung metastasis. PET/CT and CIS detected the same number of lesions in abdominal lymph nodes, hepatic metastasis, and peritoneal carcinomatosis. PET/CT detected two more metachronous tumors than did CIS in the lung and thyroid gland. Conclusion PET/CT detected more recurrences in patients who underwent surgery for CRC than did CIS and had the additional advantage of evaluating the entire body during a single scan.