Background: The standard of care for local treatment of patients (pts) with colorectal liver metastases (CRLM) is surgical resection. However, growing evidence suggests thermal ablation to be associated with a superior safety profile, lower costs, and shorter hospital stay, while rivaling surgical resection in terms of local control and overall survival (OS). This study aimed to explore the potential non-inferiority of thermal ablation compared to surgical resection for pts with small-size (≤3cm) resectable CRLM. Methods: In this multicenter, phase 3 Dutch Colorectal Cancer Group trial, pts aged 18 years and older with previously untreated CRLM were recruited from 14 centers in the Netherlands, Belgium and Italy. Pts with ≤10 CRLM, no extrahepatic metastases and ECOG 0-2 were stratified into low, intermediate and high disease burden subgroups and randomly assigned (1:1) to undergo surgical resection or thermal ablation. Though approach was left at the discretion of the operator, laparoscopic (+/- robot) resections and percutaneous ablations were favored over open procedures. To avoid drop-outs patients undergoing open procedures were randomized intra-operatively. The primary outcome was overall survival (OS) (log-rank; power 80%, 5% type I error rate; 1-sided). Secondary outcomes include distant and local tumor progression-free survival (PFS), local control, safety, length of hospital stay, quality of life and cost-effectiveness. Results: A total of 341 patients were enrolled; 299 were randomly assigned: 147 assigned to thermal ablation, 148 to surgical resection; 4 were excluded after randomization for not having the disease assessed. The trial was stopped at halftime for having met predefined stopping rules. After a median follow-up time of 28.8 months there was no difference regarding OS (HR 1.042; 95% CI, 0.689-1.576; p = 0.846) with a conditional probability of >90% to prove the hypothesis of non-inferiority. Procedure related mortality was 2.1% (n=3) for resection vs. 0% (n=0) for thermal ablation. The total number of adverse events (p = <0.001), the length of hospital stay (median 4 days [range 1-36] vs 1 day [range 1-44], p = <0.001) and local control also favored thermal ablation (HR 0.184; 95% CI, 0.040-0.838; p = 0.029). No differences were found regarding local (HR 0.833; 95% CI, 0.473-1.469; p = 0.528) and distant PFS (HR 0.982; 95% CI, 0.739-1.303; p = 0898). Conclusions: In conclusion, transitioning from surgical resection to thermal ablation as standard of care for patients with small-size (≤3 cm) CRLM would reduce complications, shorten hospital stay and improve local control, without compromising disease-free and overall survival. COLLISION is funded by a Medtronic-Covidien Investigator Sponsored Research grant. Clinical trial information: NCT03088150.