OBJECTIVE: Despite World Health Organization recommendations, concerns about promoting resistance have impeded implementation of isoniazid preventive therapy (IPT) for tuberculosis (TB). We describe characteristics of TB in individuals previously exposed to IPT as part of 'Thibela TB', a cluster-randomized trial of community-wide IPT in gold miners in South Africa. DESIGN: Case series including participants who were dispensed IPT, attended at least one follow-up visit and were subsequently treated for TB. METHODS: TB episodes were detected through surveillance and through follow-up if IPT was stopped early. Drug susceptibility data were compared with TB episodes detected through surveillance in control clusters (where IPT use was minimal) and a laboratory substudy of mycobacterial sputum culture from TB suspects in control clusters. RESULTS: Among 126 eligible individuals (125 men, median age 43 years), median time from starting IPT to TB treatment was 316 days (interquartile range 174-491). Ninety-four of the 126 (75%) were first episodes. Eighty-nine of 103 (86%) tested HIV-infected, with the median CD4 cell count of 196 cells/microl (n = 51). Sixty-four of 108 (59%) with known treatment outcomes were cured or completed treatment. Among 71 isolates with drug susceptibility results available, 12.1% [95% confidence interval (CI) 5.0-23.3] and 7.7% (95% CI 0.2-36.0) from first and retreatment episodes, respectively, had isoniazid resistance, compared with 6.0% (95% CI 3.1-10.2) and 18.7% (95% CI 10.6-29.3) in control clusters and 11.8% (95% CI 8.2-16.3) among first TB episodes in the laboratory substudy. CONCLUSION: TB after recent IPT has prevalence of drug resistance similar to background and treatment outcomes typical of this setting. These data support wider implementation of IPT.