Background: The efficacy of various screening variables in detection of diabetes mellitus (DM) is unclear. Objective: To determine the efficacy of various diagnostic tests for type 2 DM. Methods: 1021 inhabitants of Hakimieh district of Tehran aged between 30 and 75 years were studied. Known cases of diabetes and those with factors influencing glucose tolerance test were excluded. Age, sex, family history of DM, history of gestational diabetes, body mass index (BMI), waist hip ratio, blood pressure, urine glucose 2 hours after breakfast, fasting plasma glucose (FPG) and serum glucose 2 hours after consumption of 75 gm glucose (2hPG) were determined. Sensitivity, specificity and predictive values of each of these variables in comparison to WHO criteria were assessed. Results: According to WHO criteria, 5.9% of the population had undiagnosed diabetes. This rate declined to 3.8 % when employing the latest American Diabetes Association (ADA) diagnostic criteria (FPG³126 mg/dl). 88% of these newly diagnosed diabetics were diagnosed by means of 2hPG of whom 15% were classified within the normal group by ADA criteria. The 2hPG test showed higher reproducibility than the FPG test (84% vs. 67%). Truncal obesity had the maximum sensitivity (67%) and glycosuria had the highest specificity (99%). Therefore, none of these indices could be considered as a reliable screening method. Regarding 87% sensitivity and 54% specificity, a single method can be used as a proper screening test for men under 50 years of age. In women of similar age, two positive risk factors might be recruited as a useful screening method. However, the sensitivity and specificity approximated 82% and 59%, respectively. Conclusion: 2hPG is the most reliable test for diagnosis of DM. Screening via risk factors is not useful in people older than 50 years. Individuals aged <50 years, men with at least one positive risk factor and women with a minimum of two positive risk factors, are appropriate candidates for final diagnosis by glucose tolerance test.