Background: The "obesity paradox" has not been elucidated in the long-term outcomes in acute myocardial infarction (AMI) patients. This study sought to characterize the relationship between body mass index (BMI) and the risk of new-onset atrial fibrillation (NOAF). Methods: A total of 4282 participants free from AF at baseline were identified at Beijing Friendship Hospital. Baseline body mass index (BMI) was categorized into four groups. Incidence of NOAF was observed at the follow-up visits. The associations between different BMI categories and the incidence of NOAF were assessed by multivariate Cox regression analysis. Results: Over a median follow-up period of 42.0 months, 4282 participants (age 62.7 +/- 6.6 years, 38.7% women) were enrolled, 23.0% were BMI <23.0kg/m(2), 22.5% were 23.0-24.9 kg/m(2), 44.3% were 25.0-29.9 kg/m(2) and 10.2% were >30.0 kg/m(2). Compared with patients with the lowest BMI levels, those with BMI >= 30 kg/m(2) showed a younger, higher inflammatory response and a larger left atrium and were more likely to be combined with traditional cardiovascular risk factors. After adjustment for confounding variables, compared to BMI >= 30 kg/m(2) group, patients with lower BMI (<23 kg/m(2)) significantly increased the risk of NOAF in AMI patients (HR 2.884, 95% CI 1.302-6.392). Moreover, the all-cause mortality and cardiac mortality in BMI <23.0kg/m(2) group was apparently higher than that in BMI >= 30 kg/m(2) group after a long-term follow-up. Conclusion: In this AMI cohort study, the present finding of an inverse association between BMI and risk of NOAF supports the "obesity paradox". Decreasing BMI was associated with an increased risk of NOAF.