To describe tuberculosis infection among persons experiencing homelessness in inner Melbourne, Australia.

Homeless people were surveyed during late 1995 and early 1996. In stage one of the study 284 homeless people from crisis and long-term accommodation sites were recruited by means of stratified, systematic, random sampling. In stage two a convenience sample of 100 homeless people from squats and the streets were recruited. Participants completed a questionnaire and Mantoux testing was performed.

A past history of tuberculosis was reported by 3%. Thirty-seven per cent had a Mantoux > or =10 mm; 21% > or =15 mm; and 11% > or =20 mm. A Mantoux > or =15 mm was independently associated with being aged > or =40 years, coming from the accommodated sample, overseas birth, and a past history of tuberculosis. Using logistic regression modelling, a Mantoux > or =15 mm was predicted by being aged > or =40 years, overseas birth, and past history of tuberculosis.

Mantoux test results suggest that this group of homeless people had a high prevalence of infection with the tubercle bacillus. Many aspects of the physical and social circumstances of homeless people predispose to reactivation and have the potential to enhance rapid spread should latent infection become active disease.