Abstract Background The long-term efficacy and safety of combining left atrial appendage closure (LAAC) plus catheter ablation (CA) in patients with atrial fibrillation (AF) in a single procedure have been established. However, the optimal combining strategy has not been well elucidated. Objective To determine the impact of different combining strategies on long-term clinical outcomes. Methods In this prospective multicenter study, patients with non-valvular AF referred for the combining procedure were randomly assigned to the ablation-first or occlusion-first groups at a 1:1 ratio. We evaluated the outcomes of LAAC with a Watchman device and CA using a contact-force sensing catheter during long-term follow-up. Results Of the 202 randomized patients, 194(96.0%) completed the trial with 97 patients in each group. Baseline characteristics were comparable between groups. All procedures were accomplished with acute successful LAAC and restoration of sinus rhythm. Incidences of periprocedural complications were similar. Higher chronic peri-device leak (PDL, 15.5% vs 5.2%, P=0.031) and device-related thrombus (DRT, 8.2% vs 1.0%, P=0.035) rates were observed in the ablation-first group. The event-free survival rate of the primary endpoint (composite of stroke/transient ischemic attack, thromboembolism, clinically-relevant bleeding, cardiovascular rehospitalization/death) was significantly higher in the occlusion-first group (83.5% vs 71.1%, log-rank P=0.036, HR 0.53, 95%CI 0.29-0.95) during a median 2.5-year follow-up. Patients with male gender and higher CHA2DS2-VASc score were at lower risk in the subgroup analysis. Long-term freedoms from AF (77.3% vs 63.5%, log-rank P=0.039) and atrial tachyarrhythmias (70.1% vs 55.7%, log-rank P=0.044) were higher in the occlusion-first group, respectively. Conclusions The occlusion-first approach is recommended for combined procedures where a plug-like device is implanted due to its superior long-term clinical benefits.Flowchart of the COMBINATION trialIncidences of primary endpoint