The emergence of novel pathogens often has dramatic negative effects on previously unexposed host populations. Subsequent disease can drive populations and even species to extinction. After establishment in populations, pathogens can continue to affect host dynamics, influencing the success or failure of species recovery efforts. However, quantifying the effect of pathogens on host populations in the wild is challenging because individual hosts and their pathogens are difficult to observe. Here we use long-term mark-recapture data to describe the dynamics of reintroduced populations of an endangered amphibian (Rana sierrae) and evaluate the success of these recovery efforts in the presence of a recently-emerged pathogen, the amphibian chytrid fungus Batrachochytrium dendrobatidis. We find that high B. dendrobatidis infection intensities are associated with increases in frog detectability and reductions in survival. When average infection intensities are high, adults are more likely to gain infections and less likely to lose infections. We also find evidence for intensity-dependent survival, with heavily infected individuals suffering higher mortality. These results highlight the need in disease ecology for probabilistic approaches that account for uncertainty in infection intensity using imperfect observational data. Such approaches can advance the understanding of disease impacts on host population dynamics, and in the current study will improve the effectiveness of species conservation actions.