<A<em>b</em>stractText>Existing guidelines for management of type 2 dia<em>b</em>etes recommend a patient-centred approach to guide the choice of pharmacological agents. A<em>lt</em>hough glucagon-like peptide-1 (GLP-1) receptor agonists and sodium-glucose cotransporter-2 (SGLT2) inhi<em>b</em>itors are increasingly used as second-line agents, direct comparisons <em>b</em>etween these treatments are insufficient. In the SUSTAIN 8 trial, we compared the efficacy and safety of semaglutide (a GLP-1 receptor agonist) with canagliflozin (an SGLT2 inhi<em>b</em>itor) in patients with type 2 dia<em>b</em>etes.</A<em>b</em>stractText><p><div>(<em>b</em>)METHODS</<em>b</em>)</div>This was a dou<em>b</em>le-<em>b</em>lind, parallel-group, phase 3<em>b</em>, randomised controlled trial done at 111 centres in 11 countries. Eligi<em>b</em>le patients were at least 18 years old and had uncontrolled type 2 dia<em>b</em>etes (H<em>b</em>A<su<em>b</em>)1c</su<em>b</em>) 7·0-10·5% [53-91 mmol/mol]) on sta<em>b</em>le daily metformin therapy. Patients were randomly assigned (1:1) <em>b</em>y use of an interactive we<em>b</em> response system to su<em>b</em>cutaneous semaglutide 1·0 mg once weekly or oral canagliflozin 300 mg once daily. The primary endpoint was change from <em>b</em>aseline in H<em>b</em>A<su<em>b</em>)1c</su<em>b</em>), and the confirmatory secondary endpoint was change from <em>b</em>aseline in <em>b</em>odyweight, <em>b</em>oth at week 52. The primary analysis population included all randomly assigned patients, using on-treatment data collected <em>b</em>efore initiation of rescue medication. The safety analysis was done on a population that included all patients exposed to at least one dose of trial product. The trial was powered for H<em>b</em>A<su<em>b</em>)1c</su<em>b</em>) and <em>b</em>odyweight superiority under reasona<em>b</em>le assumptions. This trial is registered with ClinicalTrials.gov, NCT03136484.</p><p><div>(<em>b</em>)FINDINGS</<em>b</em>)</div>Between March 15, 2017, and Nov 16, 2018, 788 patients were randomly assigned to semaglutide 1·0 mg (394 patients) or canagliflozin 300 mg (394 patients). 739 patients completed the trial (367 in the semaglutide group and 372 in the canagliflozin group). From overall <em>b</em>aseline mean, patients receiving semaglutide had significantly greater reductions in H<em>b</em>A<su<em>b</em>)1c</su<em>b</em>) and <em>b</em>odyweight than those receiving canagliflozin (H<em>b</em>A<su<em>b</em>)1c</su<em>b</em>) estimated treatment difference [ETD] -0·49 percentage points, 95% CI -0·65 to -0·33; -5·34 mmol/mol, 95% CI -7·10 to -3·57; p&<em>lt</em>;0·0001; and <em>b</em>odyweight ETD -1·06 kg, 95% CI -1·76 to -0·36; p=0·0029). Gastrointestinal disorders, most commonly nausea, were the most frequently reported adverse events with semaglutide, occurring in 184 (47%) of 392 patients; whereas infections and infestations (defined using the Medical Dictionary for Regulatory Activities, version 21.0), most commonly urinary tract infections, occurred more frequently with canagliflozin, in 136 (35%) of 394 patients. Premature treatment discontinuation <em>b</em>ecause of adverse events occurred in 38 (10%) of 392 patients with semaglutide and in 20 (5%) of 394 patients with canagliflozin. One fatal adverse event confirmed unlikely to <em>b</em>e caused <em>b</em>y treatment occurred in the semaglutide group.</p><p><div>(<em>b</em>)INTERPRETATION</<em>b</em>)</div>Once-weekly semaglutide 1·0 mg was superior to daily canagliflozin 300 mg in reducing H<em>b</em>A<su<em>b</em>)1c</su<em>b</em>) and <em>b</em>odyweight in patients with type 2 dia<em>b</em>etes uncontrolled on metformin therapy. These outcomes might guide treatment intensification choices.</p><A<em>b</em>stractText>Novo Nordisk.</A<em>b</em>stractText>