BackgroundNo study has investigated the association between ultra-processed food (UPF) and pre-diabetes development. Furthermore, prior investigations on the association between UPF and the risk of type 2 diabetes (T2D) were primarily conducted in Europe and America, and studies in other regions are lacking. We investigated the association between ultra-processed foods and the risk of pre-diabetes and T2D in a cohort of Iranians.MethodsThis prospective study, with a sample size of 1954 for pre-diabetes and 2457 for T2D, was conducted among adults’ participants (aged ≥ 18 years) from the Tehran Lipid and Glucose Study (TLGS). We defined UPF intake using NOVA calcification as a proportion of total energy, and calculated its average intake during the follow-ups. The hazard ratios (HR) and 95% confidence intervals (95% CI) for pre-diabetes/T2D across tertiles of total UPF and per 10% of its increment were examined using Cox proportional hazards models. We also investigated the possibility of non-linear association using a restricted cubic spline regression.ResultsWe identified 766 and 256 cases of pre-diabetes and T2D, respectively, during a median follow-up of 7 years for pre-diabetes and 8.6 years for T2D. In the multivariable adjusted model, a 10% increase in total UPF intake was associated with a 12% higher risk of pre-diabetes (HR = 1.12; 95% 1.02, 1.23). The incidence of pre-diabetes was also higher in those in tertile 3 than those in tertile 1 (HR = 1.28; 95% CI = 1.07, 1.52). Following additional adjustment for diet quality, the results remained unchanged. Spline regression demonstrated a J-shaped association between UPF and the risk of pre-diabetes; the risk of pre-diabetes did not increase until UPF consumption exceeded about 24% of total energy intake. Of the individual UPF, hydrogenated fat/mayonnaise/ margarine group was related to an increased risk of pre-diabetes. The total UPF and its individual items were not associated with T2D.ConclusionsThis study found a positive, non-linear relationship between total UPF and the risk of pre-diabetes in Iranian adults. Our data could not show any significant association between UPF and T2D risk.