Atrial fibrillation (AF) in hypertrophic cardiomyopathy (HC) is associated with significant symptomatic deterioration, heart failure, and thromboembolic disease. There is a need for better mechanistic insight and improved identification of at risk patients. We used cardiovascular magnetic resonance (CMR) to assess predictors of AF in HC, in particular the role of myocardial fibrosis. Consecutive patients with HC referred for CMR 2003 to 2013 were prospectively enrolled. CMR parameters including left ventricular volumes, presence and percentage of late gadolinium enhancement in the left ventricle (%LGE) and left atrial volume index (LAVi) were measured. Overall, 377 patients were recruited (age 62 ± 14 years, 73% men). Sixty-two patients (16%) developed new-onset AF during a median follow up of 4.5 (interquartile range 2.9 to 6.0) years. Multivariable analysis revealed %LGE (hazard ratio [HR] 1.3 per 10% (confidence interval: 1.0 to 1.5; p = 0.02), LAVi (HR 1.4 per 10 mL/m 2[1.2 to 1.5; p < 0.001]), age at HC diagnosis, nonsustained ventricular tachycardia and diabetes to be independent predictors of AF. We constructed a simple risk prediction score for future AF based on the multivariable model with a Harrell's C-statistic of 0.73. In conclusion, the extent of ventricular fibrosis and LA volume independently predicted AF in patients with HC. This finding suggests a mechanistic relation between fibrosis and future AF in HC. CMR with quantification of fibrosis has incremental value over LV and LA measurements in risk stratification for AF. A risk prediction score may be used to identify patients at high risk of future AF who may benefit from more intensive rhythm monitoring and a lower threshold for oral anticoagulation.