This study aimed to determine the optimal cutoff value of waist circumference (WC) for predicting incident NAFLD. In this community-based prospective cohort study, we analyzed data from 5400 participants without NAFLD at baseline aged 40–69 years. NAFLD was defined as a NAFLD-liver fat score >−0.640. A Cox proportional hazards regression model was used to estimate the hazard ratio (HR) and 95% confidence interval (CI) for an association between body composition and NAFLD incidence. The predictive power of each body composition indicator was assessed by Harrell’s concordance index for Cox models. During a mean follow-up period of 12 years, there were 2366 new-onset NAFLD events. Compared with men with WC < 81 cm, the adjusted HR (95% CI) for incident NAFLD in those with WC ≥ 81 cm was 2.44 (2.23–2.67). Compared with women with WC < 78.5 cm, the adjusted HR (95% CI) for incident NAFLD in those with WC ≥ 78.5 cm was 2.54 (2.25–2.87). WC was the most significant risk factor for predicting incident NAFLD among body composition indicators in middle-aged and older Korean adults. The optimal WC cutoff point for predicting incident NALFD was 81 cm in men and 78.5 cm in women, which might assist in the early detection and prevention of NAFLD.