Aims Glycated haemoglobin (HbA1c) has been proposed as an alternative to the oral glucose tolerance test for diagnosing diabetes. We compared the cardiovascular risk profile of individuals identified by these two alternative methods. Methods We assessed the prevalence of cardiovascular risk factors in individuals with undiagnosed diabetes according to the World Health Organization classification or by the newly proposed HbA1c level ≥ 6.5% among 6258 participants of the Danish Inter99 study. Receiver operating curve analysis assessed the ability of fasting: 2-h plasma glucose and HbA 1c to distinguish between individuals at high and low risk of ischemic heart disease, predicted by the PRECARD program. Results Prevalence of undiagnosed diabetes was 4.1% [95% confidence interval (CI) 3.7-4.7%] by the current oral glucose tolerance test definition, whereas 6.6% (95% CI 6.0-7.2%) had diabetes by HbA1c levels. HbA1c-defined individuals were relatively older with higher proportions of men, smokers, lipid abnormalities and macro-albuminuria, but they were leaner and had lower blood pressure. HbA1c was better than fasting- and 2-h plasma glucose at distinguishing between individuals of high and low predicted risk of ischaemic heart disease; however, the difference between HbA1c and fasting- and 2-h plasma glucose was not statistically significant. Conclusions Compared with the current oral glucose tolerance test definition, more individuals were classified as having diabetes based on the HbA1c criteria. This group had as unfavourable a risk profile as those identified by the oral glucose tolerance test.