In previous studies from this laboratory, we found that approximately 30% of women with chest pain and normal coronary arteries demonstrated either a decrease in or a failure to increase radionuclide ejection fraction during exercise. To examine the hypothesis that this apparent abnormality in left ventricular function represents a physiologic difference between men and women, we prospectively studied central and peripheral cardiovascular responses to exercise in 31 age-matched healthy volunteers (16 women and 15 men). A combination of quantitative radionuclide angiography and expired-gas analysis was used to measure ejection fraction and relative changes in end-diastolic counts, stroke counts, count output, and arteriovenous oxygen difference during symptom-limited upright bicycle exercise. Normal male and female volunteers demonstrated comparable baseline left ventricular function and similar aerobic capacity, as determined by weight-adjusted peak oxygen consumption (22.1 +/- 5.1 and 22.6 +/- 4.3 ml/kg/min, respectively). However, their cardiac responses to exercise were significantly different. Ejection fraction increased from 0.62 +/- 0.09 at rest to 0.77 +/- 0.07 during exercise in men (p less than .001), but was unchanged from 0.63 +/- 0.09 at rest to 0.64 +/- 0.10 during exercise in women. The ejection fraction increased by 5 points or more in 14 of 15 men, but in only seven of the 16 women. End-diastolic counts increased by 30% in women (p less than .001), but was unchanged in men. Because decreases in ejection fraction were matched by increases in end-diastolic counts, relative increases in stroke counts and count output were the same for men and women. These data demonstrate a basic difference between men and women with respect to the mechanism by which they achieve a normal response of stroke volume to exercise; these differences must be taken into account when measurements of cardiac function during exercise stress are used for diagnostic purposes.