To achieve maximum benefit from exercise (rate)-responsive pacing in subjects with sinus node dysfunction and AV block, it is necessary to determine the pacing rate (HR) which produces maximum cardiac output (CO) under specified exercise conditions. However, the CO-HR relationship for exercise has not been systematically investigated. To permit determination of the optimum HR, CO was measured at rest and with exercise for different pacing rates. Seven dogs with complete AV block and permanently implanted ventricular pacemakers were exercised on a treadmill for 5 min at each of four pacing rates (55, 76, 101, 116/min) and at two constant exercise levels (225 and 560 kg.m/min). CO was determined by impedance cardiography during the resting state preceding exercise and during a brief period (10-20 s) immediately after exercise, and was expressed as a percent of the CO determined at rest with HR = 55/min. A three-phase pattern of CO versus HR appears to exist for exercise as for rest. For exercise, starting from a low HR, CO increases markedly; a "plateau" is reached during which moderate increase in CO is achieved by increasing HR. At very rapid pacing rates, CO may actually decrease with further increase in HR. The results of this study suggest that a subject-specific optimum HR exists for each constant exercise level. Moreover, the methodology employed in the study is applicable to the identification of optimum HR for any exercise (rate)-responsive pacemaker.