The relation between inflammation and prothrombotic state in atrial fibrillation (AF) is well recognized. This suggests a potential role for high-sensitivity C-reactive protein (hs-CRP), a marker of systemic inflammation, in improving prediction of stroke in participants with AF. Cox proportional hazard analysis was used to examine the risk of stroke in 25,841 participants (40% black and 55% women) with and without AF who were enrolled in the Reasons for Geographic and Racial Differences in Stroke study from 2003 to 2007. Baseline AF (n = 2,132) was ascertained by electrocardiogram and self-reported history of previous physician diagnosis. Stroke events were identified and adjudicated during 8.3 years of follow-up. A total of 655 incident strokes occurred during follow-up. In a model adjusted for sociodemographics, traditional stroke risk factors, and use of aspirin and warfarin, higher levels of hs-CRP were associated with increased overall stroke risk (hazard ratio [HR] 1.30, 95% confidence interval [CI] 1.10 to 1.54, and HR 1.06, 95% CI 1.01 to 1.12 for hs-CRP >3 mg/L and per 1-SD increase, respectively). Higher levels of hs-CRP continued to be associated with incident stroke in participants without AF (HR 1.31, 95% CI 1.09 to 1.57, and HR 1.06, 95% CI 1.01 to 1.12 for hs-CRP >3 mg/L and per 1-SD increase, respectively) but not in those with AF (HR 1.22, 95% CI 0.78 to 1.91, and HR 1.01, 95% CI 0.82 to 1.23 for hs-CRP >3 mg/L and per 1-SD increase, respectively). In conclusion, although hs-CRP was significantly associated with stroke risk in this population, it seems to be limited to those without AF. These findings suggest a limited value of hs-CRP in improving stroke risk stratification in subjects with AF.