In a previous study, other investigators recommended second-look surgery for colorectal cancer primarily on the basis of plasma carcinoembryonic antigen (CEA) rises and prepared a nomogram for ready recognition of these "significant" increases. We found 25 patients whose CEA levels met the recommended criteria for significance; however, in 9 of these patients the rises were transient. Eight had no clinical evidence of recurrent cancer and they might have had negative second-look surgery had this been done because of CEA rises alone. The use of the CEA nomogram merely eliminated laboratory variation as a cause of the CEA rise. It did not, however, rule out biologic causes of CEA rises, other than that of cancer, especially benign liver disease. We were unable to differentiate benign from malignant rises on the basis of CEA changes alone. Preoperative CEA values helped to separate the two rises. Transient rises usually began earlier. Malignant CEA rises were more likely to be exponential. The rate of rise alone did not discriminate between the two rises. Thus, although serial CEA levels were helpful in making the decision for reexploration, they did not substitute for complete clinical assessment.