BACKGROUND: Semaglutide is an effective treatment for type 2 diabetes; however, 20-30% of patients given semaglutide 1·0 mg do not reach glycaemic treatment goals. We aimed to investigate the efficacy and safety of once-weekly semaglutide 2·0 mg versus 1·0 mg in adults with inadequately controlled type 2 diabetes on a stable dose of metformin with or without a sulfonylurea. METHODS: We did a 40-week, randomised, active-controlled, parallel-group, double-blind, phase 3B trial (SUSTAIN FORTE) at 125 outpatient clinics in ten countries. Participants (≥18 years) with inadequately controlled type 2 diabetes (HbA1c 8·0-10·0%) with metformin and with or without sulfonylurea were randomly assigned (1:1) by an interactive web-response system to 2·0 mg or 1·0 mg once-weekly semaglutide. Participants, site personnel, the clinical study group, and investigators were masked to the randomised treatment. Outcomes included change from baseline at week 40 in HbA1c (primary outcome) and bodyweight (secondary confirmatory outcome), evaluated through trial product estimand (no treatment discontinuation or without rescue medication) and treatment policy estimand (regardless of treatment discontinuation or rescue medication) strategies. This study is registered with ClinicalTrials.gov, NCT03989232; EudraCT, 2018-004529-96; and WHO, U1111-1224-5162. FINDINGS: Between June 19 and Nov 28, 2019, of 1515 adults assessed for eligibility, 961 participants (mean age 58·0 years [SD 10·0]; 398 [41%] women) were included. Participants were randomly assigned to once-weekly semaglutide 2·0 mg (n=480 [50%]) or 1·0 mg (n=481 [50%]); 462 (96%) patients in the semaglutide 2·0 mg group and 471 (98%) in the semaglutide 1·0 mg group completed the trial. Mean baseline HbA1c was 8·9% (SD 0·6; 73·3 mmol/mol [SD 6·9]) and BMI was 34·6 kg/m2 (SD 7·0). Mean change in HbA1c from baseline at week 40 was -2·2 percentage points with semaglutide 2·0 mg and -1·9 percentage points with semaglutide 1·0 mg (estimated treatment difference [ETD] -0·23 percentage points [95% CI -0·36 to -0·11]; p=0·0003; trial product estimand) and -2·1 percentage points with semaglutide 2·0 mg and -1·9 percentage points with semaglutide 1·0 mg (ETD -0·18 percentage points [-0·31 to -0·04]; p=0·0098; treatment policy estimand). Mean change in bodyweight from baseline at week 40 was -6·9 kg with semaglutide 2·0 mg and -6·0 kg with semaglutide 1·0 mg (ETD -0·93 kg [95% CI -1·68 to -0·18]; p=0·015; trial product estimand) and -6·4 kg with semaglutide 2·0 mg and -5·6 kg with semaglutide 1·0 mg (ETD -0·77 kg [-1·55 to 0·01]; p=0·054; treatment policy estimand). Gastrointestinal disorders were the most commonly reported adverse events (163 [34%] in the 2·0 mg group and 148 [31%] in the 1·0 mg group). Serious adverse events were similar between treatment groups, reported for 21 (4%) participants given semaglutide 2·0 mg and 25 (5%) participants given semaglutide 1·0 mg. Three deaths were reported during the trial (one in the semaglutide 1·0 mg group and two in the semaglutide 2·0 mg group). INTERPRETATION: Semaglutide 2·0 mg was superior to 1·0 mg in reducing HbA1c, with additional bodyweight loss and a similar safety profile. This higher dose provides a treatment intensification option for patients with type 2 diabetes treated with semaglutide in need of additional glycaemic control. FUNDING: Novo Nordisk.