Cardiac output was monitored continuously during 4 minutes of steady-state exercise in 34 human subjects, of whom 16 were normal volunteers, 9 had heart disease with a cardiac index within normal limits during the fourth minute of exercise, and 9 had subnormal levels of cardiac output during the same period of timed exercise. The linear flow velocity in the ascending aorta was computed using the time derivative of the aortic pressure as the input to a circuit providing approximate solution to a modification of the Navier-Stokes equation. The proportionality constant necessary to express the output as volume flow was derived by comparison of the integral of the flow curves in systole with a simultaneously determined cardiac output by the Fick or dye dilution methods with the subject at rest. The cardiac output was examined each 30 seconds throughout the 4 minute period of exercise as a percentage of the level attained during the fourth minute. Striking differences between the three groups were seen 30 seconds after the start of exercise. The normal subjects had achieved 88.4 per cent of the maximal level at 30 seconds. The patients with a subnormal steady-state cardiac output had achieved only 77.9 per cent of their final value at this time while the patients with normal steady state levels were intermediate at 30 seconds. The differences in the groups were still present but diminished at one and two minutes of exercise. The output was cumulated for the entire four minute exercise period in each patient and plotted as a function of work load (i.e., 3 to 4 minute V̇ O2). Considered in this manner, a wide separation of the patients with low steady-state cardiac output and the normal subjects was achieved without overlap. These data suggest that patients with a subnormal cardiac output during the steady state phase of exercise tend to increase their cardiac outputs more slowly in response to a given exercise than do normal subjects or patients with less severe heart disease. © 1968.