Results of standard three-hour oral glucose tolerance tests (OGTT) and intravenous glucose tolerance tests (IVGTT), performed on the same subjects, were compared in an attempt to determine their value in the diagnosis of borderline diabetes. A total of 83 such tests were carried out on 81 subjects. Applying the U.S. Public Health Service point count method to the results of the OGTT test, there were 38 normals, 23 possible diabetics and 22 diabetics. A constant (K) was calculated from the glucose disappearance rate in the IVGTT curves.K disagreed with the OGTT classification to a significant extent, especially in the possible diabetic and non-diabetic groups. Also, the correlation coefficients between K and the OGTT values were not impressive. This does not mean that one test is superior to the other, only that the accuracy of either test in diagnosing early diabetes is doubtful.Technically, the IVGTT was more difficult and time-consuming, and six of the 81 subjects suffered from thrombophlebitis at the site of glucose injection.