Background: Non-colorectal cancer liver metastases (NCRLMs) represent a therapeutically challenging condition with poorly defined locoregional treatment options. This study evaluates the safety, oncological efficacy, and survival outcomes of microwave ablation (MWA) for NCRLMs in a large single-centre series. Methods: Retrospective analysis of patients undergoing MWA for NCRLM between January 2010 and December 2024 at a high-volume hepatobiliary centre. Endpoints were safety, efficacy, textbook outcome (TO) achievement, and overall survival (OS). Results: A total of 138 patients underwent 172 MWA procedures across eight primary tumour categories via a video-assisted (n = 110, 64%) or percutaneous (n = 62, 36%) approach. Major complications (Clavien–Dindo ≥ 3) occurred in 1.8% of procedures, with a median hospital stay of 2 days and a 90-day mortality of 1.7%. The median follow-up was 24.9 (10.3, 55.8) months. Complete response (CR) was achieved in 77% of procedures and TO in 54%, with incomplete response as the main driver of TO failure. The 5-year OS varied by histology, from 100% (GIST) and 80.0% (GEP-NET) to 44.7% (breast) and 0% (pancreatic adenocarcinoma). The video-assisted ablation was associated with superior OS compared to the percutaneous one (5-year OS 54.4% vs. 26.0%, p = 0.00025). In the multivariable analysis, the percutaneous approach (HR 2.44), ECOG PS ≥ 2 (HR 6.06) and a higher tumour burden score (HR 1.09) independently predicted worse OS, whereas the histological group was not independently associated with OS. Repeat MWA was the most frequent treatment for first hepatic recurrence, and 50% of patients reached no evidence of disease (NED) at last follow-up. Conclusions: MWA is a safe and repeatable locoregional treatment for NCRLMs, with outcomes shaped by tumour biology, disease burden, patient fitness and procedural approach. The high rate of liver-dominant recurrence treated with repeat MWA, combined with a final NED rate of 50%, supports MWA as a platform for iterative locoregional disease control in selected patients.