Objective: To investigate the association of body mass index （BMI） with alcoholic fatty liver disease （NAFLD）. Methods: A cohort of 725 adult subjects underwent general health check-up at our hospital in July 2008, then were followed up in 2011. The clinical data including body mass index （BMI）, blood pressure, lab testing results and liver ultrasonic findings were retrospectively analyzed. The NAFLD was diagnosed according to the guidelines for management of nonalcoholic fatty liver disease: an updated and revised edition in 2010 based on liver ultrasound results. The risk factors for NAFLD were analyzed with multivariate logistic regression. Results: One hundred and sixty two NAFLD cases and 563 non-NAFLD cases were found in 2008 check-up. Among 563 non-NAFLD subjects, NAFLD was developed in 132 （23.4%）at follow up in 2011. The incidence of NAFLD was correlated with the baseline BMI （χ²=82.861, P＜0.01）. Multivariate logistic regression analysis showed that baseline BMI, systolic blood pressure, alanine aminotransferase and the increase of BMI were the independent risk factors, while high density lipoprotein-cholesterol （HDL-C）was the protective factor for the development of NAFLD. Among 162 NAFLD cases, 71（43.8%）had no evidence of NAFLD at the second check-up in 2011. The remission of NAFLD was negatively correlated with baseline BMI （χ²=22.425, P＜0.01）. Multivariate logistic regression analysis showed that male sex, baseline BMI and the increase of BMI were negatively associated with remission of NAFLD, while the age was positively associated with the remission of NAFLD. Conclusion: The development and remission of NAFLD are frequently encountered in health check-up subjects, which are closely related to baseline BMI and changes of BMI during the follow-up.