Demographic models identify whether animals are vulnerable to local extirpation, but including all ecological parameters across life history stages may be impeded by practical difficulties. When processes acting on certain life stages cannot be measured, extrapolations are often made. A previous study documented that the range of the turtle Myuchelys georgesi is restricted to the Bellinger River, New South Wales, Australia, and its population is stable. We assessed whether M. georgesi selects certain habitats by comparing their distribution among different water holes. We assessed the threat of catfish predation by examining the stomach contents of catfish specimens. We then evaluated whether threats to M. georgesi were likely to have been underestimated by extending our previous demographic model. We did this by revising the previous estimates of adult, juvenile, and hatchling survivorship under hypothetical variations in water hole use and in the presence or absence of catfish predators. We found that M. georgesi preferentially uses moderate to deep water holes. We also found that although catfish 250−400 mm consume hatchling or juvenile turtles, those > 400 mm do to a greater extent. By making observations of catfish in the Bellinger River and incorporating their presence into our model, we found catfish presence to influence juvenile, but not adult, water hole use. Our reassessment of λ suggests that it may have been previously underestimated and that the threat to M. georgesi may be greater than we thought as the population is sensitive to variations in water hole depth and exposure of juveniles to predators. Events that alter key habitats and expose turtles to fish predators across the river should, accordingly, be evaluated further so they can be accounted for when managing the river.