Standard electrocardiographic criteria for exercise testing require near maximal exertion and fail to detect ischemia in the presence of previous infarction or conduction or repolarization abnormalities, in women, or in the presence of certain drugs. Changes in P-wave morphology have been suggested as having diagnostic utility; however, no specific criteria exist, and it is not clear which changes are most useful. This investigation evaluated the ability to detect the presence of coronary artery disease by examining changes in P-wave morphology during exercise. A group of 123 consecutive patients underwent maximum (symptom-limited) exercise nuclear stress tests. The electrocardiograms at rest, 2 minutes of exercise, 50% of maximum exercise time, maximum exercise, and 3 minutes of recovery were analyzed for the duration of the P wave in lead II, the duration of the terminal negative component of the P wave in lead V1, the amplitude of the terminal negative component of the P wave in lead V1, and the duration of the P wave in lead V5. These variables were then analyzed for their relation to the presence of perfusion defects. Of all the P-wave criteria tested, a change in amplitude of the negative component of the P wave in lead V1 from at rest to 50% of maximum exercise time of <0.025 mV was the most predictive of coronary artery disease.