The use of computed tomography scan (CT) of the abdomen and pelvis for surveillance of colorectal cancer (CRC) after primary curative therapy (PCT) remains controversial. Surveillance guidelines at Roswell Park Cancer Institute have included annual CT for the first 2 years after PCT. Isolated metastases from CRC may be amenable to surgical resection, potentially leading to a survival advantage. To assess this, a retrospective chart review of all 203 patients diagnosed with stage II or III CRC between January 1, 1990, and December 31, 1995, was conducted. First-year surveillance CT (CT-1) was performed for 146 of 203 patients and 81 of 146 patients had second-year surveillance CT (CT-2). CT was considered "directed" when at least 1 of the following prompted evaluation: suspicious symptoms or signs, rising carcinoembryonic antigen, findings from colonoscopies, chest x-rays, or laboratory tests. Otherwise, CT was considered "nondirected." Of 121 of 146 CT-1 and 63 of 81 CT-2 with nondirected CT, 7 of 121(5.8%) and 4 of 63 (6.4%) had proven recurrence, respectively. During 2 years of follow up, the estimated lower bound for detection of recurrence by nondirected CT was 11 of 121(9.1%). There were no apparent differences between the 2 groups in demographics, clinical presentation, surgical margins, treatment, tumor site, grade, or TNM stage. Surgical resectability of the metastases for directed and nondirected groups was 10 of 28 (36%) and 6 of 11 (54%), respectively. The median survival for the patients with recurrence in the directed and nondirected groups was 35 and 50 months, respectively. In conclusion, this retrospective study generates the hypothesis that CT surveillance may be of value. A prospective study, properly sized for power, is needed to answer this question.