The difficulty in interpreting the standard 12-lead electrocardiogram (ECG) due to the interference from muscle potentials produced by arm and leg motion makes it unsuitable during the exercise treadmill test. Likewise, the exercise lead placement ECG cannot substitute for the standard ECG due to significant errors in the former's diagnostic interpretation. This study compares the ECGs recorded via standard and exercise sites regarding frontal and horizontal plane axes, diagnosis and location of myocardial infarction and estimation of infarct size using the complete 54-criteria and 32-point Selvester QRS scoring system. The altered limb lead locations on the exercise ECG caused the QRS vectors to artifactually appear to be directed more inferiorly, posteriorly and rightward, producing a marked rightward mean frontal plane axis shift of +48 degrees (p less than 0.00001). No false positive or false negative anterior infarct was seen on the exercise lead placement ECG, whereas inferior and posterior infarcts were lost in 69% and 31% of patients, respectively. A false lateral infarct was seen in 19% of patients. Estimation of infarct size differed between the 2 ECG sets, with 11 patients increasing their infarct size by 3 to 9% and 14 others decreasing it by 3 to 15% on the exercise lead placement ECG. This study demonstrates that use of body torso positions for limb leads results in substantial QRS waveform variations that disqualify the exercise lead placement ECG as a "standard" recording. Such ECGs should therefore be labeled as "torso positioned" or "nonstandard" to prevent misuse for clinical and investigative purposes.