The aim of this study was to evaluate the safety and efficacy of using autologous, micro-fractured, minimally manipulated adipose tissue in patients with refractory knee osteoarthritis (OA). A total of 17 subjects (26 knees) with a median age of 72 years (range: 54-78 years) and a history of knee OA (Kellgren-Lawrence, grade of 3 or 4) underwent treatment with ultrasound-guided injection of micro-fractured adipose tissue. Micro-fractured fat was obtained using a minimal manipulation technique in a closed system (Lipogems), without the addition of enzymes or any other additives. The study subjects were clinically evaluated using the numerical pain rating scale (NPRS), the 100-point Knee Society Score (KSS) with its functional component (FXN), and the lower extremity activity scale (LEAS) at 6 weeks, 6 months, and 12 months following this procedure. When compared with baseline, significant improvements were noted in the mean values of NPRS, FXN, and LEAS at 6 weeks, 6 months, and 12 months. The mean KSS significantly improved at 6 weeks and 12 months. In particular, the average KSS score improved from 74 to 82, the FXN score improved from 65 to 76, and the LEAS score improved from 36 to 47. These values were significantly greater than the previously published minimal clinically important difference described for KSS and FXN in patients who underwent total knee arthroplasty for primary OA. No serious adverse events were reported. The injection of autologous, micro-fractured, minimally manipulated adipose tissue appears to be a safe and effective treatment option for patients with refractory, severe (grade 3 or 4) knee OA. This study demonstrated significant improvements in pain, quality of life, and function for at least 12 months in this study population. This intervention may represent a nonsurgical treatment option to avoid knee joint replacement in this population; however, further investigation is needed.