Acute myocardial infarctions (AMIs) are categorized according to presenting electrocardiography into ST-elevation (STE), non-STE, and bundle branch block AMIs. Data on the characteristics and risks of these categories originate mainly from voluntary registries or clinical trials and may be hampered by selection and information bias. This study evaluated these different categories, with the additional differentiation of non-STE AMIs into ST-depression (STD) AMIs and those with nonspecific electrocardiographic signs (no-ST) in an unselected cohort. From 1985 to 2004, all consecutive patients aged 25 to 74 years who were hospitalized with AMI at the study region's major clinic were registered prospectively. A total of 6,748 patients were identified, of whom 45.8% had STE, 14.0% STD, 32.4% no-ST, and 7.8% bundle branch block AMIs, respectively. There were substantial differences in medical history, presentation, and therapy among the AMI types. Even after adjusting for the latter factors, the odds ratios of 28-day case fatality compared with no-ST were 1.26 (95% confidence interval 1.01 to 1.59) for STE, 1.84 (95% confidence interval 1.39 to 2.44) for STD, and 3.18 (95% confidence interval 2.37 to 4.27) for bundle branch block. In conclusion, after considering in-hospital therapy, the difference between STE and no-ST was nonsignificant, whereas the case-fatality difference between no-ST and STD persisted, suggesting some other unknown underlying factors associated with STD.