Objectives: Recently, more sophisticated blood pressure (BP) measurements, such as central and ambulatory BP (ABP), have proven to be stronger predictors of future cardiovascular disease than conventional office BP. Their predictive value for atrial fibrillation development is not established. We investigated the prognostic impact for incident atrial fibrillation of office, central and ambulatory BP measurements in a predominantly older population-based cohort., Methods: Of 1004 participants in the Cardiovascular Abnormalities and Brain Lesions (CABL) study, 769 in sinus rhythm with no history of atrial fibrillation or stroke (mean age 70.5 years) underwent ABP and arterial wave reflection analysis for central BP determination. Fine and Gray's proportional subdistribution hazards models were used to assess the association of BP parameters with incident atrial fibrillation., Results: During 9.5 years, atrial fibrillation occurred in 83 participants. No office BP variable showed a significant association with incident atrial fibrillation. Central SBP and central pulse pressure were marginally associated with incident atrial fibrillation in multivariate analysis. Among ABP variables, 24-h SBP [adjusted hazard ratio per 10 mmHg, 1.24; 95% confidence interval (CI) 1.07--1.44; P = 0.004], daytime SBP (adjusted hazard ratio per 10 mmHg, 1.21; 95% CI 1.04--1.40; P = 0.011) and night-time SBP (adjusted hazard ratio per 10 mmHg, 1.22; 95% CI 1.07--1.39; P = 0.002) were significantly associated with incident atrial fibillation., Conclusion: In a predominantly older, stroke-free community-based cohort, ABP was a better independent predictor of incident atrial fibrillation than central BP, whereas office BP was inadequate for this purpose.