To determine the optimal strategy for tuberculin testing in elderly persons, we examined two aspects of the test: ( 1) the concordance between tuberculin skin test readings 2 and 7 days after application, and ( 2) the frequency of boosting of nonsignificant reactions. Four hundred eleven nursing home residents (mean age 74.5 yr) were tested. Significant reactions (⩾ 10 mm of induration to 5TU PPD) were found in 133 (32%) subjects on day 2. Men, nonwhites, and persons younger than 75 yr of age had an increased proportion of significant reactions (p < 0.01). Tuberculin reactivity steadily declined in persons between 75 and 90 yr of age but was increased among persons age 90 to 101 yr of age. Of 380 subjects with readings on both day 2 and day 7, 96 (25%) had significant reactions on both, 23 (6%) had indurations ⩾ 10 mm on day 2 only, and 20 (5%) had indurations ⩾ 10 mm on day 7 only. Repeat testing found an additional 14 subjects whose reactions increased from < 10 mm to ⩾ 10 mm (booster effect). This evaluation indicates that the most sensitive and efficient strategy for tuberculin testing consists of a single reading at day 7 followed by a second test with a reading in 48 h for persons with reactions less than 10 mm.