Electrocardiographic (ECG) Q- and ST-T–wave abnormalities predict coronary heart disease (CHD) and total mortality. No comparison has been made of the classification of these abnormalities by the 2 most widely used ECG coding systems for epidemiologic studies—the Minnesota Code (MC) and Novacode (NC). We evaluated 12-lead electrocardiograms from 64,597 participants (49 to 79 years old, 82% non-Hispanic white) in the Women's Health Initiative clinical trial in 1993 to 1998, with a maximum of 11 years of follow-up. We used MC and NC criteria to identify Q-wave, ST-segment, and T-wave abnormalities for comparison. In total, 3,322 participants (5.1%) died during an average 8-year follow-up, and 1,314 had incident CHD in the baseline cardiovascular disease-free group. Independently, ECG myocardial infarction criteria by the MC or NC were generally equivalent and were strong predictors for CHD death and total mortality (hazard ratio 1.62, 95% confidence interval 1.05 to 2.51 for CHD death; hazard ratio 1.36, 95% confidence interval 1.09 to 1.71 for total mortality) in a multivariable analytic model. Electrocardiograms with major ST-T abnormalities by the MC or NC coding system were stronger in predicting CHD deaths and total mortality than was the presence of Q waves alone. In conclusion, the ECG classification systems for myocardial infarction/ischemia abnormalities from the MC and NC are valuable and useful in clinical trials and epidemiologic studies. ST-T abnormalities are stronger predictors for CHD events and total mortality than isolated Q-wave abnormalities.