1250 protocols of heart-lung-machines, taken between 1962-1974, underwent a variance analysis. The results show that rollerpumps with high pumpvolumes (80 ccm) are not significantly lower hemolytic than rollerpumps with a pumpvolume of 40 ccm. The hemolysis stands in direct proportion to the blood flow. Membranoxygenators effect a less traumatisation than screen- and bubbleoxygenators. Furthermore the use of fresh ACD-blood shows a lower hemolysis than hemodilution and older ACD stored blood. Type and age of the vitium have no influence on the traumatic hemolysis. The use of the orthograde perfusion is significantly better than the retrograde perfusion; the use of blood-filters is no cause for traumatic hemolysis. The above mentioned results are in accordance to literature, but the facts cannot be objectivated that easily than in invitro experiments.