The association between circulating retinol-binding protein 4 (RBP4) and risk of type 2 diabetes has been inconsistent in cross-sectional studies, but prospective evidence is limited. We aimed to investigate whether plasma RBP4 is associated with future development of type 2 diabetes and whether the association could be explained by iron or other risk factors. A total of 2091 Chinese adults aged 50-70 y were followed up for 6 y. Baseline dietary intakes and fasting plasma RBP4, ferritin, adiponectin, C-reactive protein (CRP), γ-glutamyltransferase, creatinine, and erythrocyte fatty acids were determined. Self-reported doctor-diagnosed diabetes, or usage of antidiabetic agents, or fasting plasma glucose concentration at the follow-up visit ≥7.0 mmol/L was defined as an incident diabetes case. Plasma RBP4 concentration was significantly associated with dietary heme iron intake, plasma ferritin concentration, and other established risk factors. After multivariate adjustment for demographic and lifestyle variables, relative risk (RR) for type 2 diabetes when the extreme quartiles of RBP4 were compared was 1.75 (95% CI: 1.30, 2.37; P-trend < 0.001). This association remained significant when the extreme quartiles were compared (RR = 1.48; 95% CI: 1.06, 2.05; P-trend = 0.036) after further controlling for ferritin and dietary factors, as well as other risk factors, including body mass index, adiponectin, CRP, lipids, liver and kidney function, insulin resistance, and hypertension. A threshold effect of RBP4 concentrations on incident diabetes was suggested by restricted quadratic spline analysis (P = 0.026 for nonlinearity). Our study indicates that plasma RBP4 is independently associated with the 6-y risk of developing type 2 diabetes.