The genetical and medical importance of the research on the relations between blood groups and disease in the human populations requires a notable caution.
In the present paper the need is pointed out that the samples be representative, sufficient and homogeneous and some elements are noted which might introduce sampling biases.
Then the classic methods of analysis according to Aird, Woolf and Haldane are mentioned and their advantages and disadvantages outlined.
Finally a sequential method of analysis is presented that may be applied very easily and usefully by the researchers who might have the data directly and continuously in the hospitals.