Among 746 first degree relatives of diabetics we have performed oral glucose tolerance tests (GTT) very similar to the recommendations of the Committee on Statistics of the American Diabetes Association. The number of pathologic cases according to our method was 223 (or 29.9 per cent), according to the Wilkerson Point System 95 (or 12.7 per cent), according to the criteria of Fajans and Conn 213 (or 28.6 per cent) and after the University Group Diabetes Program 196 (or 26.3 per cent). For subgroups of different sex, age and body weight the relations were similar. The main difference between the four methods consists in a different sensitivity in finding pathologic test results. In addition the different methods of evaluation do not indicate the same subjects as being diabetic. 20 to 25 per cent of the diabetics with one method are nondiabetic with another despite the same over-all number of diabetics found. A fifth method is proposed which seems to be superior to all the other methods because the sum of those cases which are positive or negative only with this method is the lowest. According to this method a person is diabetic if the sum of the one-hour and the two-hour values of the GTT is 300 mg./100 ml. or more. Two hundred and forty-six (or 33.0 per cent) of our probands were diagnosed diabetic with this method. Onlyby followup studies will a decision be possible as to which method of evaluation of the GTT givesthe best results, i.e., the lowest number of false positive and false negative cases.