To compare the ability of a body shape index (ABSI) and body adiposity estimator (BAE) to BMI, waist circumference (WC) and waist:height ratio (WHtR) for predicting development of type 2 diabetes mellitus (T2DM) in rural adult Chinese. The prospective cohort study included 11 687 eligible participants who were free of T2DM at baseline. The risk of new-onset T2DM for ABSI, BAE, BMI, WC and WHtR quintiles was estimated by Cox proportional-hazards regression at follow-up. We also compared the power of ABSI and BAE to BMI, WC and WHtR for predicting the development of T2DM. With increasing ABSI, BAE, BMI, WC and WHtR, T2DM incidence was substantially increased (Ptrend<0·0001). After adjustment for multi-covariates, risk of T2DM was increased from the second to fifth quintile as compared with first quintile for ABSI (1·27; 95 % CI 0·95, 1·69; 1·35; 95 % CI 1·00, 1·82; 1·75; 95 % CI 1·33, 2·32 and 1·87; 95 % CI 1·40, 2·49; Ptrend<0·0001); BAE (1·82; 95 % CI 1·38, 2·41; 1·93; 95 % CI 1·38, 2·68; 2·73; 95 % CI 1·94, 3·84 and 4·18; 95 % CI 2·98, 5·87; Ptrend<0·0001); BMI (1·42; 95 % CI 1·03, 1·97; 1·62; 95 % CI 1·18, 2·23; 2·59; 95 % CI 1·92, 3·50 and 3·90; 95 % CI 2·90, 5·26; Ptrend<0·0001); WC (1·53; 95 % CI 1·08, 2·17; 1·66; 95 % CI 1·18, 2·33; 2·72; 1·97, 3·76 and 4·09; 95 % CI 2·97, 5·62; Ptrend<0·0001); and WHtR (1·40; 95 % CI 0·98, 1·99; 2·06; 95 % CI 1·47, 2·88; 2·90; 95 % CI 2·10, 4·01 and 4·22; 95 % CI 3·05, 5·85; Ptrend<0·0001). ABSI, BAE, BMI, WC and WHR were effective and comparable in discriminating cases from non-cases of T2DM. Risk of T2DM was increased with elevated ABSI and BAE, but the predictive ability for T2DM did not differ than that of BMI, WC and WHtR in a rural Chinese population.