The effect of estimating the blood balance using changes in erythrocyte volumes (EVs) instead of the routinely used changes in haematocrit values was studied in 20 patients scheduled for cardiac surgery. We determined the mean haematocrit of the effluent from the postoperative thoracic drainage system at various time intervals. These data were used to more accurately calculate the blood balance. From 8h after surgery onwards, the haematocrit in the thoracic effluent was less than 10%. Total loss of thoracic effluent until 24h after removal of the aortic crossclamp (ACC) was 1,735 +/- 803 ml. Calculated blood loss until 24 h after ACC was only 58% of the total thoracic effluent. Plasma volumes in these patients increased from preoperative values of 2,505 +/- 499ml at admission to the hospital to maximum levels of 4,969 +/- 1,027 ml at 12 h after ACC (p < 0.05). Blood volume rose to 159% of the preoperative value at 12 h after ACC, whereas the EV remained relatively stable, decreasing to 95% of the preoperative value at 4 h after ACC and increasing to 107% of the baseline value at 24 h after ACC. In the meantime, patient haematocrit decreased to 78% of the reference value at the time of induction of anaesthesia at 4 h after ACC and then increased to 84% at 24 h after ACC. Thus, the use of patient haematocrit considerably overestimates blood loss. The EV appears to be a more appropiate variable than haematocrit in monitoring the blood balance in cardiac surgical patients. Future studies should reveal whether the EV is practicable in daily clinical practice.