Résumé : Background and Objectives: Selection for surgery in patients with colorectal liver metastases (CRLM) remains inaccurate. We evaluated if CRLM baseline metabolic characteristics, assessed by [18]F-fluorodeoxyglucose-positron emission tomography/computed tomography (18FDG-PET/CT), could predict postoperative outcomes. Methods: In a retrospective series of patients undergoing surgery for CRLM, we defined two groups: the long-term survival (LTS) and early relapse (ER) groups, where the postoperative recurrence-free survivals were ≥5 years or <1 year, respectively. We analyzed the patients in whom baseline 18FDG-PET/CT was available. Clinicopathologic parameters, clinical risk score (CRS), and baseline 18FDG-PET/CT characteristics were compared between LTS and ER groups. A metabolic CRS (mCRS) was implemented, adding one point to the standard five-point CRS when the highest tumor standardized uptake values (SUVmax)/normal liver mean SUV (SUVmean(liver)) ratios were >4.3, defining low- and high-risk mCRS by scores of 0 to 2 and 3 to 6, respectively. Results: From a series of 450 patients operated for CRLM (mean follow-up of 58 months), we included for analysis 23 and 30 patients in the LTS and ER groups, respectively. Clinicopathologic parameters and CRS were similar in the LTS and ER groups. Median SUVmax/SUVmean(liver) ratios were higher in ER vs LTS patients (4.2 and 2.8, P =.008, respectively). mCRS was increased in ER patients (P =.024); 61% of LTS patients had low-risk mCRS and 73% of the ER patients had high-risk mCRS (P =.023). Conclusions: 18FDG-PET/CT characteristics combined with traditional CRS may represent a new tool to improve selection for surgery in patients with CRLM.