We have performed a study to assess the relative contributions of increased hospital admission rates with acute myocardial infarction and increased hospital case fatality to the excess mortality of subjects with elevated levels of glycohaemoglobin from myocardial infarction. Glycohaemoglobin levels were estimated by isoelectric focussing in 397 subjects without known diabetes mellitus admitted with myocardial infarction and compared with a control population reconstructed from a community sample of 1084 subjects without known diabetes mellitus screened in general practice. In the case-control comparison, glycohaemoglobin levels above the 90th centile were associated with relative risks of 3.1 (95% confidence interval 1.4-6.8) for admission with myocardial infarction and 5.3 (95% confidence interval 2.1-13.4) for death in hospital. Elevated glycohaemoglobin on admission was a predictor of both death and cardiac pump failure among those admitted with myocardial infarction, as was the presence of known diabetes. In those over 40 years of age, the top 1% of the glycohaemoglobin distribution contribute 4.3% of admissions and 9.6% of hospital deaths with myocardial infarction.