Background: Investigate real-world outcomes of early rhythm versus rate control in patients with recent onset atrial fibrillation. Methods: The Global Anticoagulant Registry in the FIELD-AF (GARFIELD-AF) is an international multi-centre, non-interventional prospective registry of newly diagnosed (≤6 weeks’ duration) atrial fibrillation patients at risk for stroke. Patients were stratified according to treatment initiated at baseline (≤48 days post enrolment), and outcome risks evaluated by overlap propensity weighted Cox proportional-hazards models. Results: Of 45,382 non-permanent atrial fibrillation patients, 23,858 (52.6 %) received rhythm control and 21,524 (47.4 %) rate control. Rhythm-controlled patients had lower median age (68.0 [Q1;Q3: 60.0;76.0] versus 73.0 [65.0;79.0]), fewer histories of stroke/transient ischemic attack/systemic embolism (9.4 % versus 13.0 %), and lower expected probabilities of death (median GARFIELD-AF death score 4.0 [2.3;7.5] versus 5.1 [2.8;9.2]). The two groups had the same median CHA2DS2-VASc scores (3.0 [2.0;4.0]) and similar proportions of anticoagulated patients (rhythm control: 66.0 %, rate control: 65.5 %). The propensity-score-weighted hazard ratios of rhythm vs rate control (reference) were 0.85 (95 % CI: 0.79–0.92, p-value