Background: Dapagliflozin is a sodium-glucose cotransporter-2 inhibitor approved for the treatment of type 2 diabetes. We aimed to assess the efficacy and safety of dapagliflozin as an add-on to adjustable insulin in patients with inadequately controlled type 1 diabetes. Methods: DEPICT-1 was a double-blind, randomised, parallel-controlled, three-arm, phase 3, multicentre study done at 143 sites in 17 countries. Eligible patients were aged 18-75 years and had inadequately controlled type 1 diabetes (HbA1c between ≥7·7% and ≤11·0% [≥61·0 mmol/mol and ≤97·0 mmol/mol]) and had been prescribed insulin for at least 12 months before enrolment. After an 8 week lead-in period to optimise diabetes management, patients were randomly assigned (1:1:1) using an interactive voice response system to dapagliflozin 5 mg or 10 mg once daily, given orally, or matched placebo. Randomisation was stratified by current use of continuous glucose monitoring, method of insulin administration, and baseline HbA1c. The primary efficacy outcome was the change from baseline in HbA1c after 24 weeks of treatment in the full analysis set, which consisted of all randomly assigned patients who received at least one dose of study drug. An additional 55 patients who were incorrectly and non-randomly allocated to only dapagliflozin treatment groups were included in the safety analysis set. This study was registered with ClinicalTrials.gov, number NCT02268214; data collection for the present analysis was completed on Jan 4, 2017, and a 28 week extension phase is ongoing. Findings: Between Nov 11, 2014, and April 16, 2016, 833 patients were assigned to treatment groups and included in safety analyses (dapagliflozin 5 mg [n=277] vs dapagliflozin 10 mg [n=296] vs placebo [n=260]; 778 of these patients were randomly assigned and included in the full analysis set for efficacy analyses (259 vs 259 vs 260; difference due to randomisation error affecting 55 patients). Mean baseline HbA1c was 8·53% (70 mmol/mol; SD 0·67% [7·3 mmol/mol]). At week 24, both doses of dapagliflozin significantly reduced HbA1c compared with placebo (mean difference from baseline to week 24 for dapagliflozin 5 mg vs placebo was -0·42% [95% CI -0·56 to -0·28; p<0·0001] and for dapagliflozin 10 mg vs placebo was -0·45% [-0·58 to -0·31; p<0·0001]). Among patients in the dapagliflozin 5 mg (n=277), dapagliflozin 10 mg (n=296), and placebo (n=260) groups, the most common adverse events were nasopharyngitis (38 [14%] vs 36 [12%] vs 39 [15%]), urinary tract infection (19 [7%] vs 11 [4%] vs 13 [5%]), upper respiratory tract infection (15 [5%] vs 15 [5%] vs 11 [4%]), and headache (12 [4%] vs 17 [6%] vs 11 [4%]). Hypoglycaemia occurred in 220 (79%), 235 (79%), and 207 (80%) patients in the dapagliflozin 5 mg, dapagliflozin 10 mg, and placebo groups, respectively; severe hypoglycaemia occurred in 21 (8%), 19 (6%), and 19 (7%) patients, respectively. Adjudicated definite diabetic ketoacidosis occurred in four (1%) patients in the dapagliflozin 5 mg group, five (2%) in the dapagliflozin 10 mg group, and three (1%) in the placebo group. Interpretation: Our results suggest that dapagliflozin is a promising adjunct treatment to insulin to improve glycaemic control in patients with inadequately controlled type 1 diabetes. Funding: AstraZeneca and Bristol-Myers Squibb.