In twelve French geriatric centers, a study was made of the reproducibility of the oral glucose tolerance test. The investigation included 312 persons whose ages averaged 81 years. Each patient was treated 6 times over a period of 6 months. Reproducibility of the oral glucose tolerance test is poor, the more so with increasing level of glycemia. The variability of measurements of glycemia is greater in old age than in the mid-years of adulthood. The practical conclusion is that one should be extremely cautious in diagnosing abnormal glucose tolerance in an aged person merely on the basis of a single test. Several tests in the same patient are required to achieve greater precision in glucose investigation.