The results of a detection program for diabetes over a four-year period, using a finger blood test after a carbohydrate load, were analyzed. After excluding persons with known diabetes or incomplete data, 250,938 had the initial screening test and 9,682 or 3.9 per cent screened “positive.” Approximately 75 per cent of the “positive” screenees returned for a confirmatory carbohydrate tolerance test. Five thousand, three hundred and seventy persons, or approximately 74 per cent of those who returned, had “positive” confirmatory tests. Since only persons who were “positive” in both tests were considered to have diabetes, the prevalence rate for previously undetected diabetes was 2.1 per cent. When allowance was made for “positive” screenees who failed to return by assuming that they would confirm “positive” at the same rate as the returnees, the estimated prevalence rate for previously undetected diabetes was 2.8 per cent. A representative sample of 100,077 persons showed very similar test results which were further analyzed by sex, race, weight and family history in relation to decades of age. The prevalence rate of undetected diabetes and the rate of “positive” confirmation for “positive” screenees varied markedly with age. The importance of considering the other variables in relation to age for planning detection programs is discussed; 60 per cent of the persons with diabetes detected in this program had a negative family history for diabetes. Information from return postcards sent to “positive” screenees from three months to three years after their initial test showed that a very high percentage of screenees had consulted their physicians as a result of the test; 62 per cent of the confirmed “positives” who replied stated that the diagnosis of diabetes had been confirmed by their physicians.