In 36 patients undergoing routine cardiac catheterization, a new "contact electrode" catheter technique was used to record monophasic action potentials (MAPs) from right atrial and right and left ventricular endocardial sites without the application of suction. Although of smaller amplitude, typically ranging from 15 to 40 mV, and of different reversal ratio (33 +/- 3%), MAP recordings closely resembled transmembrane action potentials in configuration and duration. Continuous MAP recordings of stable amplitude and, during regular pacing, of constant duration (+/- 1% at 90% repolarization) could be made from the same endocardial site for test periods of 1 hour (n = 4), permitting direct evaluation of the effect of cycle length alterations on local myocardial repolarization. A linear relation was found between MAP duration and basic cycle length varying from 350 to 700 ms. These rate-dependent changes in MAP duration were caused by a change in the slow phase of repolarization (phase 2), whereas the slope of rapid repolarization (phase 3) was unaltered. Single premature MAPs or MAPs after a pause showed changes in both phases. No MAPs could be recorded in areas of infarcted, aneurysmal myocardium, indicating that local viable myocardium is a prerequisite for the generation of the monophasic signal. Thus, in human subjects this catheter permits safe, long-term recording of MAPs which, although of smaller amplitude than transmembrane action potentials, bear appropriate and predictable phase relations. Such recordings may be useful in evaluating changes in local myocardial electrical activity induced by pacing or resulting from myocardial disease, or both.