The results of a population-based study involving 2 638 adult subjects residing in two districts of the Moscow region are considered in terms of the currently available screening models for detection of disturbances in carbohydrate metabolism; a new screening modality is proposed. The merits and demerits of different screening models were evaluated based on the comparison of validity of different diagnostic tests deduced from the estimation of sensitivity, specificity, and prognostic value of positive and negative results; the ROC-curve was constructed. It was shown that screening based on diagnostics of disturbances in carbohydrate metabolism from the fasting blood glucose level (FBGL) alone according to 2006 WHO diabetes criteria fails to reveal type 2 diabetes mellitus (DM2) in every fifth patient. Screening with partial exclusion of the oral glucose tolerance test (OGTT), i.e. confining it to subjects with impaired fasting glycemia (IFG) according to WHO criteria, allows up to 92.5% of DM2 cases to be diagnosed. However, this approach grossly underestimates the incidence of impaired glucose tolerance (IGT). The application of the 2003 ADA diagnostic criteria leads to a significant increase in the fraction of IFG subjects (up to 40%) while 18.2% of the patients with DM2 remain unidentified. Because the probability of diagnosis of DM2 (from 2-h point time of OGTT) and IGT increases in subjects with OGTT over 5.0 mmol/l, it appears desirable to use OGTT in the screening of cases with FBGL in excess of 5.0 mmol/l. This model may simplify the screening procedure (by avoiding OGTT) in 23.3% of those to be examined.