Frank-lead electrocardiograms (ECGs) were recorded from 59 adult males with suspected coronary artery disease. Three records were recorded consecutively on frequency modulated tape for each patient in (1) supine position, (2) sitting position with arms relaxed, and (3) sitting position on bicycle with arms on bicycle handles. Electrodes were applied at the level of the fifth intercostal space with patients in the sitting position. Computer measurements of electrocardiographic amplitudes were averaged over ten seconds of each record with these results: (1) QRS spatial amplitudes and R amplitudes in lead z were significantly higher and R amplitudes in lead y lower for sitting than for supine positions. (2) Except for slightly higher R and S amplitudes in lead x for sitting with arms on bicycle, no significant differences were observed between the two sitting positions. (3) These postural differences are significantly greater than those resulting from day-to-day variability of electrode locations. It is hypothesized that electrode level shifts with postural changes are responsible for the observed x- and z-lead changes. For the y-lead changes, it is hypothesized that shifting blood volumes with postural changes are the cause. It is concluded that reference electrocardiographic measurements for stress testing should be obtained from resting ECGs with the patient in the same postural position as that maintained during exercise.