Résumé : Background: The benefit of surgery in patients with non-colorectal non-neuroendocrine liver metastases (NCRNNELM) remains controversial. At the population level, several statistical prognostic factors and scores have been proposed but inconsistently verified. At the patient level, no selection criteria have been demonstrated to guide individual therapeutic decision making. We aimed to evaluate potential individual selection criteria to predict the benefit of surgery in patients undergoing treatment for NCRNNELM. Methods: Data for 114 patients undergoing surgery for NCRNNELM were reviewed. In this population, we identified an early relapse group (ER), defined as patients with unresectable recurrence < 1 year postoperatively who did not benefit from surgery (N = 28), and a long-term survival group (LTS), defined as patients who were recurrence-free ≥ 5 years postoperatively and benefited from surgery (N = 20). Clinicopathologic parameters, the Association Française de Chirurgie (AFC) score, and a modified 4-point Clinical Risk Score (mCRS) (excluding CEA level) were analyzed and compared between LTS and ER groups. Results: The majority of patients were female and a majority had an ASA score ≤ 2 at the time of liver surgery. The median age was 55 years. Almost half of the patients (46%) presented with a single-liver metastasis. Intermediate- and low-risk AFC scores represented 40% and 60% of the population, respectively. Five- and 10-year overall survival (OS) and disease-free survival (DFS) rates were 56% and 27%, and 30% and 12%, respectively. Negative prognostic factors were the size of liver metastases > 50 mm and delay between primary and NCRNNELM <24 months for OS and DFS, respectively. AFC score was not prognostic while high-risk mCRS (scores 3-4) was predictive for the poorer OS. The clinicopathologic parameters were similar in the ER and LTS groups, except the presence of N+ primary tumor, and the size of liver metastases was significantly higher in the ER group. Conclusion: In patients with resectable NCRNNELM, no predictive factors or scores were found to accurately preoperatively differentiate individual cases in whom surgery would be futile from those in whom surgery could be associated with a significant oncological benefit.