The achievement of glycaemic control remains challenging for patients with type 1 diabetes. We assessed the effectiveness of day-and-night hybrid closed-loop insulin delivery compared with sensor-augmented pump therapy in people with suboptimally controlled type 1 diabetes aged 6 years and older. In this open-label, multicentre, multinational, single-period, parallel randomised controlled trial, participants were recruited from diabetes outpatient clinics at four hospitals in the UK and two centres in the USA. We randomly assigned participants with type 1 diabetes aged 6 years and older treated with insulin pump and with suboptimal glycaemic control (glycated haemoglobin HbA1c 7·5–10·0%) to receive either hybrid closed-loop therapy or sensor-augmented pump therapy over 12 weeks of free living. Training on study insulin pump and continuous glucose monitoring took place over a 4-week run-in period. Eligible subjects were randomly assigned using central randomisation software. Allocation to the two study groups was unblinded, and randomisation was stratified within centre by low (<8·5%) or high (≥8·5%) HbA1c. The primary endpoint was the proportion of time that glucose concentration was within the target range of 3·9–10·0 mmol/L at 12 weeks post randomisation. Analyses of primary outcome and safety measures were done in all randomised patients. The trial is registered with ClinicalTrials.gov, number NCT02523131, and is closed to accrual. From May 12, 2016, to Nov 17, 2017, 114 individuals were screened, and 86 eligible patients were randomly assigned to receive hybrid closed-loop therapy (n=46) or sensor-augmented pump therapy (n=40; control group). The proportion of time that glucose concentration was within the target range was significantly higher in the closed-loop group (65%, SD 8) compared with the control group (54%, SD 9; mean difference in change 10·8 percentage points, 95% CI 8·2 to 13·5; p<0·0001). In the closed-loop group, HbA1c was reduced from a screening value of 8·3% (SD 0·6) to 8·0% (SD 0·6) after the 4-week run-in, and to 7·4% (SD 0·6) after the 12-week intervention period. In the control group, the HbA1c values were 8·2% (SD 0·5) at screening, 7·8% (SD 0·6) after run-in, and 7·7% (SD 0·5) after intervention; reductions in HbA1c percentages were significantly greater in the closed-loop group compared with the control group (mean difference in change 0·36%, 95% CI 0·19 to 0·53; p<0·0001). The time spent with glucose concentrations below 3·9 mmol/L (mean difference in change −0·83 percentage points, −1·40 to −0·16; p=0·0013) and above 10·0 mmol/L (mean difference in change −10·3 percentage points, −13·2 to −7·5; p<0·0001) was shorter in the closed-loop group than the control group. The coefficient of variation of sensor-measured glucose was not different between interventions (mean difference in change −0·4%, 95% CI −1·4% to 0·7%; p=0·50). Similarly, total daily insulin dose was not different (mean difference in change 0·031 U/kg per day, 95% CI −0·005 to 0·067; p=0·09) and bodyweight did not differ (mean difference in change 0·68 kg, 95% CI −0·34 to 1·69; p=0·19). No severe hypoglycaemia occurred. One diabetic ketoacidosis occurred in the closed-loop group due to infusion set failure. Two participants in each study group had significant hyperglycaemia, and there were 13 other adverse events in the closed-loop group and three in the control group. Hybrid closed-loop insulin delivery improves glucose control while reducing the risk of hypoglycaemia across a wide age range in patients with suboptimally controlled type 1 diabetes. JDRF, NIHR, and Wellcome Trust.