The association between ABO blood groups and certain diseases of the upper intestinal tract, peptic ulcer, gastric carinoma, and pernicious anaemia, is now well established, but this is not true for any of the other associations noted, except perhaps diabetes. The evidence has recently been summarized by Roberts (1957). Of the other associations that have been reported, the one which, if true, would be of most interest as a selective force acting on the relative frequency of the ABO groups is the associa tion between bronchopneumonia in children and the ABO groups other than O. The observation that children of blood group O are relatively resistant to bronchopneumonia was made in 1949 (Struthers 1951). It was based on 400 consecutive post-mortem examinations at the Royal Hospital for Sick Children, Glasgow. Struthers's findings were that 55 children had no apparent abnormality at death apart from broncho pneumonia, and that in this group only fourteen (25-5 per cent) were group O, compared with 51 per cent, group O children in a large control sample from the same area (x2 for one degree of freedom: 13*18). Of 93 children with bronchopneumonia associated with other abnormality, only 33 (35-5 per cent.) were group O, giving %2 for one degree of free dom: 8-16 in comparison with the contact sample. We note that there is no significant difference in the percentage of group O between the group with bronchopneumonia alone and that with broncho pneumonia in association with other abnormality, and in the combined series of 148 there are 47 group O children (31 -8 per cent.), giving %% f?r one degree of freedom: 20-5 in comparison with the control group. Struthers found that the deficit of group O children in his total series of 400 was largely confined