The term "embolic stroke of undetermined source" (ESUS) encompasses a substantial but heterogeneous population of patients with ischemic stroke, underscoring the importance of identifying personalized treatment strategies. In subgroups of patients randomized in ESUS trials, we evaluated the effectiveness of anticoagulation compared with antiplatelet therapy in secondary ischemic stroke prevention. A study-level meta-analysis was conducted on randomized controlled trials of patients with ESUS, comparing anticoagulation with antiplatelet therapy. The primary efficacy outcome was recurrent ischemic stroke, and safety outcomes were major bleeding and death. Subgroups assessed were age, sex, presence of patent foramen ovale (PFO), left atrial enlargement (LAE), and atrial cardiopathy. Pooled risk ratios (RRs) were meta-analyzed. Cochrane Risk of Bias Tool 2.0 was used for risk-of-bias assessment. A total of 7 randomized controlled trials involving 14,804 patients were analyzed, with 7,406 patients treated with anticoagulation and 7,398 treated with antiplatelet therapy. Compared with antiplatelet therapy, anticoagulation was associated with a similar rate of recurrent ischemic stroke (RR 0.91, 95% CI 0.80-1.05; I2 = 0%). In ESUS with PFO, anticoagulation was associated with significantly lower risk of ischemic stroke (RR 0.59, 95% CI 0.35-0.98; I2 = 0%). Heterogeneity was present in those with LAE: antiplatelet therapy was superior in trials allowing cardiac monitoring after randomization (RR 6.65, 95% CI 1.26-35.08; I2 = 0%), but anticoagulation was superior in trials prohibiting cardiac monitoring after randomization (RR 0.25 95% CI 0.07-0.89). Subgroups based on age, sex, or presence of atrial cardiopathy did not benefit from anticoagulation over antiplatelet therapy. In this meta-analysis, an empiric anticoagulation approach is not beneficial for patients with ESUS. This finding highlights the importance of an individualized treatment strategy. Such a strategy should include prolonged cardiac monitoring for atrial fibrillation, particularly in patients with moderate-to-severe LAE. Anticoagulation treatment showed promise in patients with medically treated PFO. Other subgroups did not benefit from anticoagulation therapy. Large prospective studies within ESUS subgroups are needed to validate our findings.