In order to discover possible differences among various ethnic groups in Hawaii with respect to the prevalence of diabetes mellitus, blood samples were secured from 38,103 gainfully employed persons. The sample, taken from a finger 2½ hr after a meal containing at least 50 gm of carbohydrate, was tested by an instrument which automatically determined whether the concentration of glucose was above or below 130 mg/100 cc. Those whose blood glucose was found to be elevated were referred to their personal physicians for further diagnostic study. The over-all, age-adjusted prevalence of diabetes was 18.4/1,000. The most striking ethnic difference was that the rate for unmixed Hawaiians was about six times that for persons classified as Caucasian. In addition to finding ethnic variations, the study served as a case-finding procedure, for three of five of those diagnosed as diabetic were previously unknown.