Background: The combination of insulin and glucagon-like peptide-1 (GLP-1) receptor agonist therapy improves glycaemic control, induces weight loss, and reduces insulin dose needed in type 2 diabetes. We assessed the efficacy and safety of the GLP-1 receptor agonist liraglutide as an add-on therapy to insulin for overweight adult patients with type 1 diabetes. Methods: We did a randomised, double-blind, placebo-controlled trial at Steno Diabetes Center (Gentofte, Denmark). Patients aged 18 years or older with type 1 diabetes, insufficient glycaemic control (HbA1c >8% [64 mmol/mol]), and overweight (BMI >25 kg/m2) were randomly assigned (1:1) to receive insulin treatment plus either liraglutide or placebo (saline solution) by subcutaneous injection once per day. Randomisation was done in blocks of four. Treatment assignment was masked to investigators and patients. Treatment lasted 24 weeks and liraglutide was started at a dose of 0·6 mg per day, escalated to 1·2 mg per day after 1 week, and then again to 1·8 mg per day after another week. Intervals between dose increments could be extended at the discretion of the investigator. The primary endpoint was change in HbA1c from baseline to week 24. Secondary endpoints were changes in hypoglycaemic events, glycaemic variability, glycaemic excursions, insulin dose, bodyweight, postprandial plasma concentrations of glucagon and GLP-1, gastric emptying, blood pressure, heart rate, patient-reported outcome measures, time spent in hypoglycaemia, near-normoglycaemia, and hyperglycaemia, plasma fasting glucose, mean glucose, and cholesterol. Efficacy analyses were calculated by use of a mixed model, whereby a patient's data are used as long as the patient is in the study. The safety analyses were done in the intention-to-treat population, which consisted of all patients who received at least one dose of their randomly assigned study drug. This study is registered with ClinicalTrials.gov, number NCT01612468. Findings: Between July 10, 2012, and May 30, 2014, we enrolled 100 patients with type 1 diabetes, with 50 patients allocated liraglutide and 50 to placebo. Four patients from the liraglutide group and six patients from the placebo group discontinued treatment before 24 weeks. At the end of treatment, change in HbA1c from baseline did not differ between groups (-0·5%, 95% CI -0·8 to -0·4 [-6·0 mmol/mol, 95% CI -8·7 to -4·4] with liraglutide vs -0·3%, -0·6 to -0·2 [-4·0 mmol/mol, -6·6 to -2·3] with placebo; between-group difference -0·2% [-0·5 to 0·1; 2·2 mmol/mol, -5·5 to 1·1], p=0·1833). The number of hypoglycaemic events was reduced with liraglutide, with an incident rate ratio of 0·82 (95% CI 0·74 to 0·90). However, we detected no changes in glycaemic variability (continuous overall net glycaemic action per 60 min from 10·3 [95% CI 9·8 to 10·8] to 9·9 [9·2 to 10·6] in the liraglutide treated patients vs 10·2 [9·7 to 10·7] to 9·7 [9·1 to 10·3] in the placebo treated patients). Both bolus insulin (difference -5·8 IU, 95% CI -10·7 to -0·8, p=0·0227) and bodyweight (difference -6·8 kg, 95% CI -12·2 to -1·4, p=0·0145) decreased with liraglutide treatment compared with placebo. Heart rate increased with liraglutide, with a difference between groups of 7·5 bpm (95% CI 2·8-12·2, p=0·0019). Postprandial plasma glucagon and GLP-1 concentrations did not differ between groups (difference between groups at end of treatment: -408 mmol/L per 240 min [95% CI -941 to 125, p=0·1309] for glucagon and -266 mmol/L per 240 min [-1034 to 501, p=0·4899] for GLP-1). Gastric emptying was delayed after 3 weeks of treatment with liraglutide (19·9 min, 95% CI 0·8 to 39·0, p=0·0412), but we detected no difference after 24 weeks of treatment (-1·5 min, -20·5 to 17·6, p=0·8793). Patient-reported outcome measures differed between groups only with respect to perceived frequency of hypoglycaemia, which was higher with placebo, with a difference between groups of -0·6 (95% CI -1·1 to -0·07, p=0·0257). Liraglutide was associated with more frequent nausea (29 [58%] patients with liraglutide vs five [10%] with placebo), dyspepsia (11 [22%] patients with liraglutide vs one [2%] with placebo), diarrhoea (ten [20%] patients with liraglutide vs one [2%] with placebo), decreased appetite (seven patients [14%] with liraglutide vs none with placebo), and vomiting (seven [14%] patients with liraglutide vs one [2%] with placebo). Interpretation: In patients with type 1 diabetes, overweight, and insufficient glycaemic control, the reduction in HbA1c did not differ between insulin plus placebo and insulin plus liraglutide treatment. Liraglutide was associated with reductions in hypoglycaemic events, bolus and total insulin dose, and bodyweight, and increased heart rate. Funding: Novo Nordisk.