Poison baiting is used frequently to reduce the impacts of pest species of mammals on agricultural and biodiversity interests. However, baiting may not be appropriate if non-target species are at risk of poisoning. Here we use a desktop decision tree approach to assess the risks to non-target vertebrate species in Australia that arise from using poison baits developed to control feral house cats (Felis catus). These baits are presented in the form of sausages with toxicant implanted in the bait medium within an acid-soluble polymer capsule (hard shell delivery vehicle, or HSDV) that disintegrates after ingestion. Using criteria based on body size, diet and feeding behaviour, we assessed 221 of Australia's 3,769 native vertebrate species as likely to consume cat-baits, with 47 of these likely to ingest implanted HSDVs too. Carnivorous marsupials were judged most likely to consume both the baits and HSDVs, with some large-bodied and ground-active birds and reptiles also consuming them. If criteria were relaxed, a further 269 species were assessed as possibly able to consume baits and 343 as possibly able to consume HSDVs; most of these consumers were birds. One threatened species, the Tasmanian devil (Sarcophilus harrisii) was judged as definitely able to consume baits with implanted HSDVs, whereas five threatened species of birds and 21 species of threatened mammals were rated as possible consumers. Amphibia were not considered to be at risk. We conclude that most species of native Australian vertebrates would not consume surface-laid baits during feral cat control programs, and that significantly fewer would be exposed to poisoning if HSDVs were employed. However, risks to susceptible species should be quantified in field or pen trials prior to the implementation of a control program, and minimized further by applying baits at times and in places where non-target species have little access.