Abstract Introduction/Objective To determine the effectiveness of carcinoembryonic antigen (CEA) in patient with known residual disease after resection of colonic cancer Methods Serial carcinoembryonic antigen (CEA) levels were measured in 43 patients following incomplete excision of colorectal cancer. Twenty-seven patients received chemotherapy. The patients reported in this study were amongst those participating in a randomized trial of chemotherapy following palliative resection of colorectal cancer. All patients had a resection of colorectal cancer leaving behind macroscopic quantities of tumour, though of a small total residual bulk. The residual sites of tumour were recorded as: (a) local - where there was local extension of the tumour; (b) nodal - where there were clinically involved nodes remaining, with histological confirmation of involved nodes in the resected specimen; (c) hepatic - where small hepatic metastases were noted, individually not greater than 2 cm in diameter and in total representing less than 10% of the liver bulk; (d) peritoneal - where peritoneal tumour seedlings remained, with histological confirmation of at least one of the seedlings. Results Of the 43 patients studied after operation and prior to chemotherapy, the mean log CEA was 1.19 and only 3 patients exceeded 1.48, despite the fact that all patients were known to have residual tumour. Similarly, the mean log CEA at 3 months for the total group remained at 1.26 but 6 patients now had abnormal values. Thereafter the mean values of the total group exceeded the upper limit of normality and increasing numbers of patients showed elevated CEA values. About one half of the patients had levels of CEA well within the normal range during the first postoperative year despite the evidence of residual disease documented at operation. Conclusion The most appropriate application for a marker such as CEA may be as an assessment of the effect of chemotherapy in patients who have a definite, but objectively unmeasurable, tumour load.