BACKGROUND: Tuberculosis is the leading single-pathogen cause of death worldwide, and China has the third largest number of cases worldwide. New tools, such as new vaccines, are needed to meet WHO tuberculosis goals. Tuberculosis vaccine development strategies mostly target infants or adolescents, but given China's ageing epidemic, vaccinating older people might be important. We modelled the potential impact of new tuberculosis vaccines in China targeting adolescents (15-19 years) or older adults (60-64 years) with varying vaccine characteristics to inform strategic vaccine development. METHODS: A Mycobacterium tuberculosis transmission model was calibrated to age-stratified demographic and epidemiological data from China. Varying scenarios of vaccine implementation (age targeting [adolescents or older adults] and coverage [30% or 70%]) and characteristics (efficacy [40%, 60%, or 80%], duration of protection [10 years or 20 years], and host infection status required for efficacy [pre-infection, post-infection in latency, post-infection in latency or recovered, or pre-infection and post-infection]) were assessed. Primary outcomes were tuberculosis incidence and mortality rate reduction in 2050 in each vaccine scenario compared with the baseline (no new vaccine) scenario and cumulative number needed to vaccinate (NNV) per case or death averted, 2025-50. FINDINGS: By 2050, results suggest that 74·5% (uncertainty interval [UI] 70·2-78·6) of incident tuberculosis cases in China would occur in people aged 65 years or older, and 75·1% (66·8-80·7) of all cases would be due to reactivation, rather than new infection. All vaccine profiles delivered to older adults had higher population-level impact (reduction of incidence and mortality rates) and lower NNV per case and per death averted than if delivered to adolescents. For an intermediate vaccine scenario of 60% efficacy, 10-year protection, and 70% coverage, the reduction of tuberculosis incidence rates with older adult vaccination was 1·9 times (UI 1·5-2·6) to 157·5 times (119·3-225·6) greater than with adolescent vaccination, and the NNV was 0·011 times (0·008-0·014) to 0·796 times (0·632-0·970) lower. Furthermore, with older adult vaccination, post-infection vaccines provided substantially greater mortality and incidence rate reductions than pre-infection vaccines. INTERPRETATION: Adolescent-targeted tuberculosis vaccines, the focus of many development plans, would have only a small impact in ageing, reactivation-driven epidemics such as those in China. Instead, an efficacious post-infection vaccine delivered to older adults will be crucial to maximise population-level impact in this setting and would provide an important contribution towards achieving WHO goals. Older adults should be included in tuberculosis vaccine clinical development and implementation planning. FUNDING: Aeras and UK MRC.