To investigate the association between the presence of ipsilateral nonstenotic carotid plaques and the rate of detection of atrial fibrillation (AF) during follow-up in patients with embolic strokes of undetermined source (ESUS). We pooled data of all consecutive ESUS patients from 3 prospective stroke registries. Multivariate stepwise regression assessed the association between the presence of nonstenotic carotid plaques and AF detection. The 10-year cumulative probabilities of AF detection were estimated by the Kaplan-Meier product limit method. Among 777 patients followed for 2,642 patient-years, 341 (38.6%) patients had an ipsilateral nonstenotic carotid plaque. AF was detected in 112 (14.4%) patients in the overall population during follow-up. The overall rate of AF detection was 8.5% in patients with nonstenotic carotid plaques (2.9% per 100 patient-years) and 19.0% in patients without (5.0% per 100 patient-years) (unadjusted hazard ratio [HR] 0.56, 95% confidence interval [CI] 0.37-0.84). The presence of ipsilateral nonstenotic carotid plaques was associated with lower probability for AF detection (adjusted HR 0.57, 95% CI 0.34-0.96, p = 0.03). The 10-year cumulative probability of AF detection was lower in patients with ipsilateral nonstenotic carotid plaques compared to those without (34.5%, 95% CI 21.8-47.2 vs 49.0%, 95% CI 40.4-57.6 respectively, log-rank-test: 11.8, p = 0.001). AF is less frequently detected in ESUS patients with nonstenotic carotid plaques compared to those without. NCT02766205.