Statistical models of epidemiology in wildlife populations usually consider diseased individuals as a single class, despite knowledge that infections progress through states of severity. Bovine tuberculosis (bTB) is a serious zoonotic disease threatening the UK livestock industry, but we have limited understanding of key epidemiological processes in its wildlife reservoirs. We estimated differential survival, force of infection and progression in disease states in a population of Eurasian badgers (Meles meles), naturally infected with bTB. Our state-dependent models overturn prevailing categorizations of badger disease states, and find novel evidence for early onset of disease-induced mortality in male but not female badgers. Males also have higher risk of infection and more rapid disease progression which, coupled with state-dependent increases in mortality, could promote sex biases in the risk of transmission to cattle. Our results reveal hidden complexities in wildlife disease epidemiology, with implications for the management of TB and other zoonotic diseases.