Background: Radiofrequency ablation (RFA) as an alternative to surgery in patients with colorectal liver metastasis (CRLM) is associated with high local recurrence (LR) rates. RFA requires a wide margin to prevent LR. It is yet unclear how large the margin should be and how this can be measured reliably. The aim of our study was to evaluate the feasibility of 3D margin assessment and correlate the obtained margins to LR rates. Methods: Patients treated with percutaneous RFA for a solitary CRLM were included. Pre- and post-ablation CT scans were loaded into Mirada RTx (Mirada Medical, Oxford, UK) and merged based on venous structures in proximity of the tumor. Tumor and ablation volumes were determined on both scans based on automatic contour detection. Overlay of the pre- and post-scans allowed 3D assessment of ablation margins. The minimal margin was compared with the occurrence of LR during follow-up. Results: Ten patients were included so far (median follow-up 35 months). LR occurred in 5 patients (median tumor size 17mm, range 11-42) in which ablation was incomplete in 4 patients. Location of the LR was at the margin most-at-risk in all patients. Five patients did not develop LR and their tumors were ablated completely (margins: 1, 2 and ≥5mm (n=3)). Median tumor size was 19 mm, range 10-27. Conclusion: Automated 3D margin assessment is feasible and appears to aid in LR prediction. Real-time margin assessment can directly influence the precision and completeness of RFA, potentially achieving improved local control.