OBJECTIVES: We describe isoniazid-related adverse events in Thibela TB, a cluster-randomized study of community-wide isoniazid preventive therapy (IPT) among gold miners in South Africa, where HIV prevalence is estimated at 30%. METHODS: Consenting employees were screened prior to IPT for active tuberculosis and increased risk of isoniazid toxicity using a questionnaire and chest radiograph. Study-defined IPT-related adverse events were sought at each study visit: liver function tests were only performed if clinically indicated. In a substudy, we questioned consecutive participants at baseline and months 1, 3, and 6 concerning minor IPT-related adverse events. RESULTS: Among 24,221 participants (95.2% men, median age 40 years), 130 individuals had 132 study-defined adverse events (0.54%); 61 (0.25%) possible hypersensitivity rash, 50 (0.21%) peripheral neuropathy, 17 (0.07%) clinical hepatotoxicity, and four (0.02%) convulsions. Four events (two hepatotoxicity, one fatal, and two convulsions) fulfilled criteria for seriousness. Clinical hepatotoxicity was associated with consumption of alcohol [0.11 vs. 0.03% if no alcohol consumed, odds ratio 3.9 (95% confidence interval 1.2-12.1)], but not with sex, age, weight, or concurrent antiretroviral therapy. In the substudy, 324 of 498 (65.1%) participants reported better health since starting IPT; 180 of 324 (55.6%) reported that this was because of increased appetite. The frequency of specific minor symptoms was low among those taking IPT, and all symptoms were reported less often than at baseline. CONCLUSION: The risk of adverse events, particularly hepatotoxicity, was very low in this population. Our data suggest that clinical criteria can safely be used for screening prior to and monitoring during IPT.