In randomized placebo-controlled trials in Haïti, Zambia and Uganda, prophylactic use of isoniazid (INH) for 6 to 12 months reduced the annual incidence of tuberculosis in HIV-infected patients by more than 50 per cent. For several years, WHO, IUTATLD and CDC have recommended that HIV-positive patients testing positive in a PPD test should be treated with INH as a form of anti-tuberculosis chemoprophylaxis (ATC). Whilst these recommendations are easy to follow in industrialized countries, widespread use of ATC in developing countries remains problematic because: (i) It is unknown what proportion of patients are likely to be re-infected at the end of ATC in countries where TB is endemic; (ii) It is possible that resistant bacilli may be selected due to the incomplete exclusion from the ATC program of patients with active TB at enrollment; (iii) It is difficult to identify asymptomatic carriers of M. tuberculosis at enrollment; (iv) It is doubtful that all patients will comply with a treatment regime which lasts several months; (v) The cost of a widespread ATC program, whose full benefit remains to be evaluated, may be difficult to justify. This paper attempts to review these issues and demonstrates the need for more population-based clinical trials in the field.