Background: Isoniazid preventive therapy (IPT) reduces tuberculosis (TB) incidence in high risk populations (HIV-positive and negative), but durability of protection is limited in high burden settings. The underlying mechanism (reactivation of persistent latent TB or re-infection) is not well understood. There are few data on long-term TB incidence after IPT discontinuation. This study aimed to investigate the timing of post-IPT TB incidence and its associated risk factors in a high-TB-risk population with high HIV prevalence (estimated at 29%) in sub-Saharan Africa. Methodology: Data from a cluster randomized trial of community-wide IPT among gold mine workers in South Africa (the ‘Thibela TB’ study) were analysed. Participants in the intervention arm who were prescribed isoniazid at least once and who were alive, in the workforce and TB-free 9 months later (corresponding to the intended duration of IPT) were included. TB incidence rates from the end of the intended period of IPT to the end of the follow-up period for the main study were calculated, overall and stratified by follow-up time (per 6 months). Cox regression analysis was used to investigate risk factors for TB incidence and effect modification by follow-up time. A sensitivity analysis used the actual end of IPT (based on prescription dates) as the start of the risk period. Results: 17,446 participants (96% male, mean age 41 years [yrs; standard deviation 9]) were eligible for the analysis. 541 TB cases were diagnosed during 24,037 person-years at risk (pyar), a median follow-up of 1.4 years. The post-IPT TB incidence rate was 2.3/100 pyar (95% confidence interval [CI], 2.1-2.5) and did not vary during follow-up time (0-0.5 yrs: 2.0/100 pyar [95% CI 1.7-2.3], 0.5-1 yrs: 2.4 [95% CI 2.1-2.8], 1-1.5 yrs: 2.5 [95% CI 2.1-3.0] and ≥1.5 yrs: 2.1 [95% CI 1.6-1.5], P-value 0.12). After adjusting for sex, country of origin and ART use, risk factors identified were older age (30-39 versus ≤29 yrs: adjusted hazard ratio [aHR] 2.2 [95% CI 1.9-4.2], 40-49 yrs: aHR 2.8 [95% CI 1.9-4.2], ≥50 yrs: aHR 2.8 [95% CI 1.8-4.2]), underground mining work (aHR 1.8 [95% CI 1.2-2.5]), living in a mining hostel (aHR of non-hostel versus hostel 0.7 [95% CI 0.6-0.9]) and prior TB disease (aHR 1.9 [95% CI 1.5-2.4]). These factors did not vary importantly over time. The results of the sensitivity analysis did not differ. Conclusions: TB incidence rates were constant over time since IPT cessation. The risk factors identified are well known and are associated with both reactivation and re-infection disease. No important effect modification of these factors by time was found. These findings do not allow conclusions regarding the mechanism underlying post-IPT TB incidence.