Objective: The identification of individuals at risk of atrial fibrillation (AF) remains a challenge. We investigated whether high-sensitive cardiac troponin I (hs-cTnI) is an independent predictor of incident atrial fibrillation (AF) hospitalisation in an Australian community-based cohort. Methods: Serum hs-cTnI was measured in 1641 men and 2189 women in the Busselton Health Study 1994/1995 Cohort aged 25–84 years at baseline. Data on incident AF hospitalisation over 20 years follow-up were obtained by data linkage. Results: Hs-cTnI was detectable (>1.2 ng/L) in 65.5% of participants (males 81.4%, females 53.6%) at baseline. There were 179 (10.9%) AF events in men and 208 (9.5%) in women. In the multivariable-adjusted model, hs-cTnI was a significant predictor of AF event with hazard ratio 1.21 (95% CI 1.11–1.32, P < 0.001) in the whole cohort, 1.17 (95% CI 1.01–1.35, P = 0.037) in men and 1.22 (95% CI 1.08–1.37, P = 0.001) in women for a doubling of baseline hs-cTnI. In women, a graded and significant increase in risk of AF was observed across hs-cTnI categories from ≥1.3 ng/L, whereas in men only the highest category (≥6.0 ng/L) had significantly increased risk compared with individuals with hs-cTnI ≤1.2 ng/L. Addition of categorical hs-cTnI to standard risk factors for AF improved risk estimation in females (C-statistic increment 0.006, P = 0.04) but not males (increment 0.005, P = 0.12) and net reclassification improvement was not significant in either sex. Conclusions: High-sensitive cTnI level is an independent predictor of incident AF hospitalisation in a community-based cohort but does not improve risk stratification.