Purpose: Laparoscopic cholecystectomy remains the standard of care for acute cholecystitis; however, a subset of patients is unsuitable for surgery due to frailty, significant comorbidity, hostile anatomy, or advanced malignancy. Endoscopic ultrasound-guided gallbladder drainage (EUS-GBD) using a lumen-apposing metal stent offers a minimally invasive alternative. This study reports clinical outcomes of EUS-GBD and characterises the preoperative factors influencing patient selection. Methodology: A retrospective single-centre audit was performed of all EUS-GBD procedures undertaken between 2020 and 2024 at a tertiary hospital in Melbourne, Australia. All procedures were performed by experienced endoscopists using lumen-apposing metal stents (AXIOS™). Demographic, clinical, and outcome data were collected. Preoperative risk was assessed using the Clinical Frailty Scale (CFS), Charlson Comorbidity Index (CCI), Karnofsky Performance Scale, and the ACS NSQIP surgical risk calculator. Results: Thirty-five patients (median age 75 years) underwent EUS-GBD. Technical success was achieved in 33/35 patients (94%), with clinical symptom resolution in 30/35 (86%). Ninety-day mortality was 26% (n=9); all deaths occurred in patients with metastatic or locally advanced malignancy. Only one patient (3%) required readmission for recurrent acute cholecystitis within 90 days. The majority of patients had active malignancy (20/35). Median CFS was 5 (IQR 4–5), median CCI was 6 (IQR 2–8), and median Karnofsky score was 70 (IQR 55–90). Sixty per cent had a NSQIP-predicted mortality exceeding 10% for laparoscopic cholecystectomy. Patients with higher functional status were selected primarily due to hostile anatomy. Conclusion: EUS-GBD is a safe and effective alternative for high-risk patients with acute cholecystitis, demonstrating high technical and clinical success with low recurrence. Current surgical risk tools broadly align with clinician judgement in patient selection; however, prospective trial data is required to help develop a dedicated framework.