Background: Insulin degludec (degludec) is a second-generation basal insulin with an improved pharmacokinetic–pharmacodynamic profile compared with first-generation basal insulins, but there are few data regarding its use during pregnancy. In this non-inferiority trial, we aimed to compare the efficacy and safety of degludec with insulin detemir (detemir), both in combination with insulin aspart (aspart), in pregnant women with type 1 diabetes. Methods: This open-label, multinational, randomised, controlled, non-inferiority trial (EXPECT) was conducted at 56 sites (hospitals and medical centres) in 14 countries. Women aged at least 18 years with type 1 diabetes who were between gestational age 8 weeks (+0 days) and 13 weeks (+6 days) or planned to become pregnant were randomly assigned (1:1), via an interactive web response system, to degludec (100 U/mL) once daily or detemir (100 U/mL) once or twice daily, both with mealtime insulin aspart (100 U/mL), all via subcutaneous injection. Participants who were pregnant received the trial drug at randomisation, throughout pregnancy and until 28 days post-delivery (end of treatment). Participants not pregnant at randomisation initiated the trial drug before conception. The primary endpoint was the last planned HbA_1c measurement before delivery (non-inferiority margin of 0·4% for degludec vs detemir). Secondary endpoints included efficacy, maternal safety, and pregnancy outcomes. The primary endpoint was assessed in all randomly assigned participants who were pregnant during the trial. Safety was assessed in all randomly assigned participants who were pregnant during the trial and exposed to at least one dose of trial drug. This study is registered with ClinicalTrials.gov, NCT03377699, and is now completed. Findings: Between Nov 22, 2017, and Nov 8, 2019, from 296 women screened, 225 women were randomly assigned to degludec (n=111) or detemir (n=114). Mean HbA_1c at pregnancy baseline was 6·6% (SD 0·6%; approximately 49 mmol/mol; SD 7 mmol/mol) in the degludec group and 6·5% (0·8%; approximately 48 mmol/mol; 9 mmol/mol) in the detemir group. Mean last planned HbA_1c measurement before delivery was 6·2% (SE 0·07%; approximately 45 mmol/mol; SE 0·8 mmol/mol) in the degludec group and 6·3% (SE 0·07%; approximately 46 mmol/mol; SE 0·8 mmol/mol) in the detemir group (estimated treatment difference −0·11% [95% CI −0·31 to 0·08]; −1·2 mmol/mol [95% CI: −3·4 to 0·9]; p_{non-inferiority}<0·0001), confirming non-inferiority. Compared with detemir, no additional safety issues were observed with degludec. Interpretation: In pregnant women with type 1 diabetes, degludec was found to be non-inferior to detemir. Funding: Novo Nordisk.