Objectives of this study were to determine the prevalence of positive tuberculin reactivity and associated factors among elderly nursing home residents in a population with a relatively high tuberculosis notification rate, to estimate the prevalence of active tuberculosis, and to assess tuberculin reactivity as a screening tool. A stratified, disproportional, randomized cluster sample of residents was selected and the Mantoux test (using 0.1 ml of 5 tuberculin units of purified protein derivative of tuberculin) carried out. All subjects with a positive test had a chest X-ray followed by sputum smear and culture if the X-ray was abnormal. Sputum examination was also carried out in a random sample of controls, matched for age and gender, drawn from subjects with a negative Mantoux test. Information regarding medical history, tobacco smoking habits, symptoms related to tuberculosis, and communal eating habits were gathered. Also anthropometric data were collected. Sixteen nursing homes in the catchment area of a major district hospital in Hong Kong comprising 587 residents (136 men, 451 women, mean age 80 +/- 8 years) participated in this study. The weighted prevalence of tuberculin reactivity was 43.8%. It was higher in men, among those who took their meals in a common area, in the younger age group, and in those with no previous history of hospitalization. No association was found between prevalence and duration of residence, smoking, skinfold thickness, past medical history, or any relevant symptoms. Following radiological and sputum examination, the estimated prevalence of active tuberculosis ranged from 1.2 to 2.6%. The sensitivity of the tuberculin test was 86, the specificity 30%. The prevalence of active tuberculosis in nursing homes in Hong Kong is high, but it is unclear whether cross-infection or poor health of the residents is the major contributing factor. Tuberculin skin testing does not appear to be a useful screening method in this population.