Objective To investigate the association between heart rate recovery (HRR) and new-onset atrial fibrillation (AF) in middle-aged adults. Patients and Methods Heart rate recovery was calculated using the exercise stress test in 15,729 apparently healthy self-referred men and women who attended periodic health screening examinations between January 2000, and December 2015. All participants completed the maximal exercise stress test according to the Bruce protocol and were followed clinically on a yearly basis for a median of 6.4±4 years. The primary end point was new-onset AF. Participants were grouped according to HRR at 5 minutes, dichotomized at the median value (<73 beats/min). Results Participants with low HRR were older, were more commonly men, had a higher rate of comorbidities, and were less fit. Kaplan-Meier survival analysis revealed that the cumulative probability of AF at 6 years was higher in participants with low HRR (2.1%) than in those with high HRR (0.6%) (log-rank, P<.001). Older age, male sex, obesity resting heart rate, and ischemic heart disease were all associated with increased AF risk in a univariate Cox regression model (P<.05 for all). Multivariate Cox regression analysis revealed that low HRR was independently associated with increased AF risk (hazard ratio, 1.92; 95% CI, 1.3-2.8; P<.001) after adjustment for multiple confounders. Conclusion Lower HRR is independently associated with the development of new-onset AF during long-term follow-up in middle-aged adults.