Aims In 2010, the American Diabetes Association has published recommendations on the population to be screened for dysglycaemia; the diagnostic criteria for intermediate hyperglycaemia and diabetes using oral glucose tolerance testing and HbA(1c); and the patients eligible for treatment with metformin. We aimed to evaluate the consequences of screening with oral glucose tolerance test or HbA(1c) in an at-risk population. Methods Among 1177 overweight or obese consecutive adults without known diabetes who were referred to our department for weight management, we selected 1157 individuals (83% female; 80% European) fulfilling the American Diabetes Association 2010 criteria for dysglycaemia screening. Results Mean age was 41.2 +/- 13 years, BMI 37.0 +/- 7.2 kg/m(2), fasting plasma glucose 4.9 +/- 0.8 mmol/l and HbA(1c) (turbidimetric immunoassay) 5.7 +/- 0.7% (39 mmol/mol). Based on oral glucose tolerance test and HbA(1c), respectively, 76 (6.6%) and 113 (9.8%) patients had diabetes, including 34 sharing both criteria; 307 (26.5%) and 478 (41.3%) had intermediate hyperglycaemia; and 130 (11.2%) and 255 (22.0%) would be treated with metformin. The sensitivity/specificity of HbA(1c) >= 6.5% (48 mmol/mol) for the diagnosis of diabetes according to the oral glucose tolerance test were 44.7/92.7%. Diabetes risk scores and UK Prospective Diabetes Study cardiovascular risk score were the highest in the 130 patients having both an abnormal oral glucose tolerance test and HbA(1c) >= 5.7%. Conclusions In a population at risk for diabetes, the HbA(1c) strategy could lead to diagnosing more cases of dysglycaemia and to treating more patients with metformin than the oral glucose tolerance test strategy. The consistency of either diagnostic criteria was low. The patients with the highest a priori risk of diabetes and cardiovascular disease were those fulfilling both oral glucose tolerance test and HbA(1c) criteria. Diabet. Med. 28, 567-574 (2011)