We assessed the prevalence and 3-month change in glucose tolerance status in consecutive non-ST-elevation myocardial infarction (NSTEMI; European Society of Cardiology 2007 definition) patients (N=49; mean (S.D.) age 65 (11) years) admitted to a coronary care unit, without known diabetes. These patients underwent an oral glucose tolerance test (OGTT) 36-hour (median, IQR: 18-72) after admission and at 3 months. Undiagnosed abnormal glucose tolerance (AGT: impaired fasting glucose (IFG), impaired glucose tolerance (IGT) or new diabetes) was common (61% at admission and 41% at 3 months, p<0.05) and the majority (approximately 3/4) had IGT. Glucose tolerance status improved in a higher proportion of patients than it worsened (31% vs. 8%, p=0.04). At 3 months, fasting glucose was unchanged but 2-hour OGTT glucose was lower (mean (S.D.): 8.5 (2.7) mmol/L vs. 7.7 (2.7) mmol/L, p=0.004). 'Stress hyperglycaemia' could explain higher admission glucose levels and this raises the question about the optimal timing of OGTT in relation to myocardial infarction. Newly diagnosed diabetes was present in approximately 10% of patients and this was not reliably detected by fasting plasma glucose. In NSTEMI patients OGTT is the only reliable strategy to identify subjects with IGT and diabetes.