A comparison was made between cardiac output values determined by the dye dilution and electrical impedance methods in ten subjects at rest and during graded exercise on a bicycle ergometer. The cardiac output values determined by the two methods were linearly related and significantly (P less than 0.001) correlated (r = 0.90). Movement artifact associated with exercise at maximum or near-maximum work loads caused severe distortion of the dZ/dt wave form and prevented calculation of impedance cardiac output at these levels of work. Use of the lowest value of L (distance between mean value of L in the impedance stroke volume equation (SV = p(L2/ZO2) (dZ/dt)mt), resulted in nearly identical values for the least-squares line and equalvalue line of impedance and dye cardiac outputs. Although absolute values of cardiac output determined by the two methods were not identical in all subjects the changes in cardiac output were nearly identical during the different levels of exercise. The data support the validity of the impedance method as a noninvasive, atraumatic measure of cardiac output at rest and during graded exercise.