Ten thousand bloods were examined for grouping by the open macroscopic method of Vincent. Forty-five and six-tenths per cent were Group O; 36.4%, Group A; 13.5%, Group B; and 4.5%, Group AB. The material used for testing these bloods consisted of pooled specimens of Group A and B sera. The grouping of the red cells was without fault in 9,985 bloods. Fifteen gave questionable agglutination because they were either clotted or very anemic. The sera of a small number were tested with known red cells and it was found that in 6 cases the typing did not check with results obtained by typing the red cells. In 5 cases, in which the donor and the recipient were of the same group, there was agglutination when the bloods were cross-matched. In 1467 blood transfusions there were but 2 deaths: one as a result of an error in technic and the 2nd as a result of use of a so-called universal donor. Ten thousand two hundred and forty-two blood transfusions collected from the literature showed 22 deaths from all causes; 13 were due to wrong types, 1 to primary disease, 4 to heart failure and 4 to unknown causes. The open macroscopic method of Vincent is efficient for determining the blood group of recipients and donors. Testing sera should consist of pooled specimens of Group A and B sera. Grouping by this method should be determined by testing the red cells alone. The classification of bloods into 4 groups is sufficient for the purpose of blood transfusion. The death rate for blood transfusions is 0.39 per thousand.[long dash]Author''s summary and conclusions.