A major unmet medical need to better manage Type 2 Diabetes (T2D) is the accurate disease prediction in subjects who show glucose dysmetabolism, but are not yet diagnosed as diabetic. We investigated the possibility to predict/monitor the progression to T2D in these subjects by retrospectively quantifying blood circulating microRNAs in plasma of subjects with i) normal glucose tolerance (NGT, n = 9); ii) impaired glucose tolerance (IGT, n = 9), divided into non-progressors (NP, n = 5) and progressors (P, n = 4) based on subsequent diabetes occurrence, and iii) newly diagnosed T2D (n = 9). We found that impaired glucose tolerance associated with a global increase of plasma circulating microRNAs. While miR-148 and miR-222 were specifically modulated in diabetic subjects and correlated with parameters of glucose tolerance, the most accentuated microRNA dysregulation was found in NP IGT subjects, with increased level of miR-122, miR-99 and decreased level of let-7d, miR-18a, miR-18b, miR-23a, miR-27a, miR-28 and miR-30d in comparison with either NGT or T2D. Interestingly, several of these microRNAs significantly correlated with parameters of cholesterol metabolism. In conclusion, we observed the major perturbation of plasma circulating microRNA in NP pre-diabetic subjects and identified a unique microRNA profile that may become helpful in predicting diabetic development.