A series of two 100 gin. and two 50 gm. oral glucose tolerance tests were performed on each of ninety-six young, healthy, prison inmates under rigidly controlled experimental conditions. Mean blood glucose levels for fasting, one, two, and three hour-postchallenge determinations were compared between challenges. The interrelationships of these determinations were also compared between the two standard loading doses.Mean two-hour levels were 14 to 27 per cent higher after the larger challenge. With an appropriate conversion factor, the absolute difference in two-hour values could be equated. Thus, despite these unequal means, the diagnostic information obtained with either challenge at one and two hours does not differ. The fact that the one-hour determination was not related to loading dose size or obesity as expressed by the weight index [100 (weight)/height 2 ] and was highly correlated with the Tolerance Index, suggests that it is the preferable screening test for diabetes. Studies in other populations are needed to confirm this, however.Although mean levels at three hours were the same, the distribution of tolerance responses for this determination differed for the two challenges. An excess of both relative hypoglycemic and hyperglycemic responses occurred with the larger loading dose. These contrasting distributions suggest that the 100 gm., three-hour determination may possess unique diagnostic power to ascertain early chemical diabetes. Until adequate criteria of an abnormal three-hour response are established, however, this determination cannot be used as a screening test for diabetes.Physiologic variability for all specific determinations and the complete tolerance test were similar for both challenges. The fact that only a moderate degree of correlation was found between repeat tolerance responses greatly compromises the clinical values of a single glucose tolerance test.