BACKGROUND: Preoperative carcinoembryonic antigen (CEA) has been reported as a prognostic factor in patients with colorectal liver metastasis (CRLM) after hepatectomy. However, the impact of a preoperative "CEA uptrend" on prognosis after hepatectomy in these patients remains unknown. This study assessed the impact of CEA uptrend on prognosis in patients undergoing hepatectomy for CRLM. METHODS: Consecutive patients with CRLM who underwent hepatectomy between 2009 and 2018 were retrospectively analyzed. Patients with CRLM for whom CEA was measured both around 1 month before (CEA-1m) and within 3 days (CEA-3d) before hepatectomy were enrolled. A CEA-3d higher than both the upper limit of normal (5 ng/ml) and CEA-1m was defined as a CEA uptrend. RESULTS: Study participants comprised 212 patients with CRLM. Of these, 88 patients (41.5%) showed a CEA uptrend. CEA uptrend indicated better discriminatory ability (corrected Akaike information criteria, 733.72) and homogeneity (likelihood ratio chi-square value, 18.80) than CEA-3d or CEA-1m. Patients with CEA uptrend showed poorer overall survival than those without CEA uptrend (p < 0.001). After adjusting for known prognostic factors, the prognostic significance of CEA uptrend retained (hazard ratio 2.63, 95% confidence interval 1.63-4.26, p < 0.001). In subgroup analyses, the prognostic significance of CEA uptrend was retained irrespective of the status of RAS mutation or response to preoperative chemotherapy. CONCLUSIONS: CEA uptrend offers better prediction of survival outcomes than conventional CEA measurements in patients undergoing hepatectomy for CRLM.