The study aimed to assess the association between total, direct, and indirect bilirubin and nonalcoholic fatty live disease (NAFLD) risk given its high prevalence and serious clinical prognosis. Among 27,009 subjects who participated in a healthy screening program from the Dongfeng-Tongji cohort study in 2008, 8189 eligible subjects (aged 35-86 years; males, 43.95%) were ultimately enrolled. The incidence rates of NAFLD in 2013 were compared with respect to baseline bilirubin levels among subjects free of NAFLD, and the effect sizes were estimated by logistic regression analysis. During 5 years follow-up, we observed 1956 cases of newly developed NAFLD with the overall incidence of 23.88%. Direct bilirubin was presented to inversely associate with NAFLD risk. Compared with quartile 1 of direct bilirubin, the multivariable-adjusted ORs (95% CIs) for NAFLD of quartile 2 to 4 were 1.104 (0.867-1.187), 0.843 (0.719-0.989), and 0.768 (0.652-0.905), respectively, P for trend 0.002). Similarly, inverse effects of direct bilirubin on NAFLD incidence were also observed when stratified by sex and BMI. However, no significant associations were found between total, and indirect bilirubin and NAFLD risk. Direct bilirubin reduced NAFLD risk independent of possible confounders among middle-aged and elderly Chinese population, probably based on the endogenous antioxidation of bilirubin.