Each of 529 persons identified as positive in a diabetes mass screening industrial program was matched with a negative screenee of the same sex, race, age, and place of employment. A confirming glucose tolerance test classified the positive screenees into three groups: confirmed (204), not confirmed (141), and not tested (184). The confirmed group was slightly older than the other two groups. A three-year follow-up by means of a mail questionnaire showed that death occurred more frequently in confirmed positives (7 per cent); the self-perceived, current health state was poorer for the confirmed positives than for their controls, with no differences between the other positive and control groups; and significant weight loss occurred in confirmed and not tested positives. A positive screening test resulted in increased early contact with physicians, but only confirmed positives had evidence of continuing increased contact. Oral antidiabetic medications were currently prescribed for 28, 17, and 1 per cent, respectively, for the confirmed, the not tested, and the not confirmed positives. With no clear evidence of modification of the subjects' behavior toward continued medical care and, considering the potential harm of improper or early exhibition of oral drugs or the incorrect labeling of persons as diabetic, the possible benefits of early physician intervention and weight loss do not appear to offset the potential nonbenefit of mass screening. These conclusions have resulted in modification of the Cleveland program to one of discriminate screening rather than indiscriminate mass screening.