Inflammation has been implicated in the pathogenesis of cardiovascular diseases. C-reactive protein (CRP), a marker of systemic inflammation, predicts the risk of coronary events and stroke. Atrial fibrillation (AF) is associated with atrial structural changes that may have an inflammatory basis. We tested the hypothesis that CRP is a risk factor for AF. Subjects were those included in the database registry of the Intermountain Heart Collaborative Study from 1994 to 2001. Patients who had >or=1 electrocardiogram that demonstrated AF formed the disease group (n = 347), and those who had neither electrocardiographic nor clinical evidence for AF comprised the control group (n = 2,449). Logistic regression assessed the quartile (Q) of CRP and 13 other clinical and angiographic predictors of AF. Average age was 63 +/- 12 years, 33% were women, and 61% had advanced coronary artery disease. Patients who had AF were older (by 7 years) and more frequently had a history of heart failure than did controls (41% vs 9%). CRP was higher in patients who had AF than in controls (p <0.001). Q-CRP was a univariable predictor of AF (odds ratio 1.39/Q, 95% confidence interval 1.25 to 1.55, p <0.001). Adjusting for age and heart failure decreased the predictive value of Q-CRP to 1.20/Q (95% confidence 1.07 to 1.34, p = 0.002), whereas further adjustment for 11 other variables had little additional effect (odds ratio 1.19/Q, 95% confidence interval 1.06 to 1.33, p = 0.003). Thus, high levels of CRP independently predicted an increased risk of AF among a large, prospectively studied patient cohort that was assessed angiographically. Increased CRP is a new risk marker for AF propensity, and testing therapies that target inflammation should be considered.