96 Background: Previous data on surgery for colorectal cancer (CRC) liver metastases (CRLM) documented improvement in overall survival (OS) and low perioperative mortality, but patient (pt) selection requires complex multidisciplinary work. Predictive scores such as the Comprehensive Evaluation of Relapse Risk (CERR) estimate relapse risk after surgery and help in the pre-operative decision. Objectives: This study aims to evaluate OS and progression-free survival (PFS) in a population of pts who had surgery for CRLM, and to analyse survival outcomes by risk groups, according to the CERR score. We identified individual, anatomic and biological factors with potential prognostic impact. Methods: We retrospectively analysed the medical records of pts who had CRLM surgery between January 2012 and December 2021 in two Portuguese reference centres. Outcomes for CRLM surgery were analysed by risk groups according to the CERR score of each pt (based on nodal status, preoperative CEA/CA 19.9 values, KRAS/BRAF status, presence of extrahepatic disease and modified tumour burden score). Univariate and multivariate survival analyses were performed with the log-rank test and Cox proportional-hazards regression models. Results: A total of 174 pts were included: 66,1% (n=115) male, mean age at CRLM diagnosis 62,1 (±11,5) years, 80,5% (n=140) with isolated liver metastases. Thirty-three pts (19%) had a second surgery, independently of the CERR score (chi-square p = 0,60). Eighteen (10,3%) pts had a CERR score of 0-1 (low-risk, LR), 96 (55,2%) of 2-3 (medium-risk, MR) and 60 (34,5%) of 4-5 (high-risk, HiR). With a median (MED) follow-up of 85,1 (15,6-186,5) months (mo), 100 (57.5%) pts have died. MED OS was 88,0 mo (95% confidence interval [CI] 40,6-135,4) for LR, 46,4 mo (95% CI 38,7-54,1) for MR and 30,7 mo (95% CI 23,1-38,4) for HiR pts (p=0,02); ECOG performance status was an independent prognostic factor (p=0,002) for OS. Median PFS was 88,0 mo (95% CI 6,9-169,0) for LR, 12,9 mo (95% CI 9,1-16,7) for MR and 7,7 mo (95% CI 5,9-9,4) for HiR pts (p<0,001); positive resection margins and prior induction chemotherapy (CT) were additional independent negative prognostic factors for PFS (p<0,001). Conclusions: These results suggest a significant association between relapse risk after surgery for CRLM and both OS and PFS. The CERR score can be used to select potential candidates for CRLM surgery and improve the impact of this intervention in pts with CRC. The link between induction CT and worse survival may reflect selection bias as these pts possibly had a higher volume of disease and therefore received initial systemic therapy. Prospective studies are required to confirm these results.