Abstract: Aims: We analysed the Finnish Diabetes Prevention Study data in order to evaluate how the new HbA_1c-based criterion compares with the oral glucose tolerance test in diagnosing Type 2 diabetes among high-risk individuals during a prospective average follow-up of 4 years. Methods: In the Diabetes Prevention Study, 172 men and 350 women who were overweight and had impaired glucose tolerance at baseline were randomized into an intensive lifestyle intervention or a control group. The oral glucose tolerance test and HbA_1c measurements were performed annually until the diagnosis of diabetes using the World Health Organization 1985 criteria. Results: The sensitivity of the HbA_1c≥ 6.5% (≥ 48 mmol/mol) as a diagnostic criterion for Type 2 diabetes was 35% (95% CI 24%, 47%) in women and 47% (95% CI 31%, 64%) in men compared with diagnosis based on two consecutive oral glucose tolerance tests. The corresponding sensitivities for HbA_1c≥ 6.0% (≥ 42 mmol/mol) were 67% (95% CI 55%, 77%) and 68% (95% CI 51%, 82%). The participants with HbA_1c≥ 6.5% (≥ 48 mmol/mol) and diabetes based on the oral glucose tolerance test were more obese and had higher fasting glucose and 2-h glucose concentrations than those who had a diabetic oral glucose tolerance test but HbA_1c < 6.5% (< 48 mmol/mol). There were no differences in the predictive performance of baseline fasting glucose, oral glucose tolerance test and HbA_1c. Conclusions: Of those with diabetes diagnosis based on two oral glucose tolerance tests during the Diabetes Prevention Study follow-up, 60% would have remained undiagnosed if diagnosis had been based on HbA_1c≥ 6.5% (≥ 48 mmol/mol) criterion.