SETTING: Human immunodeficiency virus (HIV) clinics at two Thai tertiary care medical centres. OBJECTIVES: To evaluate the efficacy of tuberculin skin test (TST) guided isoniazid preventive therapy (IPT) in combination with antiretroviral therapy (ART) in the prevention of tuberculosis (TB). DESIGN: A 4-year prospective comparative study of patients at two HIV clinics: one performed TST at enrolment and, if positive, prescribed IPT (IPT group), while the other did not perform TST (non-IPT group). RESULTS: There were 200 patients included in each group. Baseline characteristics and drop-out rates were similar in both groups. The incidence of pulmonary TB over 4 years was not significantly different between the IPT and non-IPT groups (0.80 cases vs. 1.76 per 100 person-years [py], P = 0.13). However, the incidence of pulmonary TB in the non-IPT group was significantly higher during the first 6 months (8.60 vs. 0 cases/100 py, P = 0.01) and among patients with initial CD4 < 200 cells/μl (9.41 vs. 0 cases/100 py, P = 0.02). The survival analyses demonstrated a protective effect of IPT (χ 2 = 3.66, P = 0.04) for early TB. CONCLUSIONS: Benefit of IPT plus ART was evident only in the first 6 months of care. These findings suggest that TST-guided IPT should be routinely provided for HIV-infected patients after initial entry into medical care. © 2012 The Union.