Before there is widespread clinical application of the high-frequency ECG, differences resulting from the leads used and the measurement criteria for late potentials must be resolved. Therefore 113 consecutive patients without resting QRS conduction abnormalities referred for Holter monitoring were studied. Four different lead systems were used: a standard bipolar orthogonal lead system and three bipolar lead systems mapping the left ventricle. Measurements made of late potentials included normal and high-frequency QRS duration, their difference, the duration of low-amplitude signals (less than 40 uV) in the terminal QRS, and the root mean square of the last 40 msec of the high-frequency QRS duration. We found that the left ventricular leads tended to give more abnormal measurements than the orthogonal system and that the various measurements failed to agree with each other. In addition, even in this population in which abnormalities of QRS conduction were excluded, the late potential measurements tended to be more abnormal as QRS duration lengthened. These differences in lead systems and measurement criteria must be considered when clinically applying information regarding late potentials measured from the high-frequency ECG.