Widespread prey depletion forces carnivores to rely more on livestock, which may lead to increased persecution by humans. Reliable quantification of livestock consumption is essential for understanding depredation scales, but a comparative analysis of extant biomass models used for this purpose has never been done before. We conducted a global meta-analysis of two linear and three non-linear biomass models used to estimate consumption of prey biomass and individuals by seven big cat species. We applied the z-test to perform pairwise comparisons of estimates produced by five models for each prey record. Further, we used logistic regression to assess the effects of species of big cats and their prey, scat sample size, prey body mass, and study sites on significantly different and similar estimates. The analysis of 769 prey records from 47 sites demonstrated that, in over 95% of cases, linear and non-linear biomass models produced similar estimates of prey biomass and individuals consumed. Significantly different estimates of prey biomass consumed (in 1.5% of cases) and prey individuals consumed (4%) were obtained only in certain study sites and for a few big cat species (tiger Panthera tigris, leopard Panthera pardus, and puma Puma concolor). Due to the paucity of different estimates, the effects of predictors could not be ascertained. Our study demonstrated that linear models tend to estimate higher biomass of large prey, lower biomass of medium-sized prey and fewer individuals of large and medium-sized prey consumed than non-linear models. This disagreement in estimates suggests that the numbers of livestock lost to depredation can be underestimated by linear models, and that re-calculation by non-linear models is required. However, the difference between estimates produced by linear and non-linear models is generally small and such re-calculation may be recommended only for tiger, puma and leopard in certain areas.