Simultaneous Fick and duplicate dye cardiac outputs were done in 105 patients with various cardiovascular diseases during routine cardiac catheterization. Dye was injected into the pulmonary artery and sampled from the brachial artery. Nineteen patients had mitral and/or aortic valvular regurgitation. Eighty-four per cent of the duplicate dye cardiac outputs agreed within 10 per cent variation from the line of identity, and 98 per cent were within 25 per cent. There was no systematic difference between the Fick and dye methods. Seventy-five per cent agree within 20 per cent variation from the line of identity. However, individual variation ranged from -27 to +58 per cent. There was, also, no systematic difference between Fick and dye methods either with low cardiac index or valvular regurgitation. Variation between the two methods was less with low cardiac index and greater with higher cardiac index. The variation was not increased in the presence of valvular regurgitation. The variation in the two methods could partly be explained by errors in the measurement of arteriovenous oxygen difference and oxygen consumption. When the injection is made into the pulmonary artery and sampled from the brachial artery, dye outputs are valid irrespective of the level of resting cardiac index and valvular regurgitation as long as there are enough points to draw a straight line from semilogarithmic trace of the descending limb.