Policy development, implementation, and effective contingency response rely on a strong evidence base to ensure success and cost-effectiveness. Where this includes preventing the establishment or spread of zoonotic or veterinary diseases infecting companion cats and dogs, descriptions of the structure and density of the populations of these pets are useful. Similarly, such descriptions may help in supporting diverse fields of study such as; evidence-based veterinary practice, veterinary epidemiology, public health and ecology. As well as maps of where pets are, estimates of how many may rarely, or never, be seen by veterinarians and might not be appropriately managed in the event of a disease outbreak are also important. Unfortunately both sources of evidence are absent from the scientific and regulatory literatures. We make this first estimate of the structure and density of pet populations by using the most recent national population estimates of cats and dogs across Great Britain and subdividing these spatially, and categorically across ownership classes. For the spatial model we used the location and size of veterinary practises across GB to predict the local density of pets, using client travel time to define catchments around practises, and combined this with residential address data to estimate the rate of ownership. For the estimates of pets which may provoke problems in managing a veterinary or zoonotic disease we reviewed the literature and defined a comprehensive suite of ownership classes for cats and dogs, collated estimates of the sub-populations for each ownership class as well as their rates of interaction and produced a coherent scaled description of the structure of the national population. The predicted density of pets varied substantially, with the lowest densities in rural areas, and the highest in the centres of large cities where each species could exceed 2500 animals.km-2. Conversely, the number of pets per household showed the opposite relationship. Both qualitative and quantitative validation support key assumptions in the model structure and suggest the model is useful at predicting the populations of cats at geographical scales important for decision-making, although it also indicates where further research may improve model performance. In the event of an animal health crisis, it appears that almost all dogs could be brought under control rapidly. For cats, a substantial and unknown number might never be bought under control and would be less likely to receive veterinary support to facilitate surveillance and disease management; we estimate this to be at least 1.5 million cats. In addition, the lack of spare capacity to care for unowned cats in welfare organisations suggests that any increase in their rate of acquisition of cats, or any decrease in the rate of re-homing might provoke problems during a period of crisis.