A hybrid closed-loop (HCL) insulin delivery system was evaluated to establish its safety for unsupervised use in patients =14 years. The system included the Medtronic MiniMed 670G pump, 4th-generation sensors, and a control algorithm. Patients calibrated the sensor periodically and gave mealtime and correction boluses as needed. A 2-wk run-in (baseline) phase was followed by a 3-mo study phase of HCL at home and supervised hotel settings for 5 nights followed by an optional continued-access program. Data were available from 124 patients with T1D (55 male) with mean (±SD) age, 37.8±16.46 yr (30 age =21) and duration of diabetes 21.7±13.65 yr. Sensor glucose (SG) and A1c values from baseline and study phases were compared. HCL mode was used for a median 87.2% (IQR, 75.0% to 91.7%) of the time after first start. There were higher percentages of SG 71-180 mg/dL, lower percentages of SG =70 mg/dL, and lower percentages of SG =50 mg/dL during 24 hours and at night (p<.001 for each) in the study phase compared to baseline (Table). Mean A1c decreased from 7.4±0.9% to 6.9±0.6% (p<.001). SG variability measured by coefficient of variation decreased from 0.38 to 0.35 (p<.001). There was no DKA, severe hypoglycemia, or serious device-related adverse event during 12,389 patient-days. At study’s end, 99 patients entered the continued-access program. The HCL system was safe, acceptable, and associated with improved glucose control during extended at-home use.