Flavonoid-rich foods have shown a beneficial effect against non-alcoholic fatty liver disease (NAFLD) in short-term randomised trials. It is uncertain whether the usual dietary intake of flavonoids may benefit patients with NAFLD. The present study evaluated the association between the usual intake of flavonoids and the risk of progression in NAFLD. The prospective study included 2694 adults from the Guangzhou Nutrition and Health Study. Face-to-face interviews using a seventy-nine-item FFQ were administered to assess habitual dietary flavonoid intake, while abdominal ultrasonography was conducted to evaluate the presence and degree of NAFLD, with measurements conducted 3 years apart. After adjustment for potential confounders, higher flavonoid intakes were gradely associated with reduced risks of worsen NAFLD status. The relative risks of worsening (v. non-worsening) NAFLD in the highest (v. lowest) quintile were 0·71 (95 % CI 0·54, 0·93) for total flavonoids, 0·74 (95 % CI 0·57, 0·95) for flavanones, 0·74 (95 % CI 0·56, 0·96) for flavan-3-ols, 0·90 (95 % CI 0·68, 1·18) for flavonols, 0·73 (95 % CI 0·56, 0·93) for flavones, 0·79 (95 % CI 0·61, 1·02) for isoflavones and 0·74 (95 % CI 0·57, 0·96) for anthocyanins. An L-shaped relationship was observed between total flavonoid intake and the risk of NAFLD progression. Path analyses showed that the association between flavonoids and NAFLD progression was mediated by decreases in serum cholesterol and homeostasis model assessment of insulin resistance. This prospective study showed that higher flavonoid intake was associated with a lower risk of NAFLD progression in the elderly overweight/obese Chinese population.