Stable isotopes are useful tools for estimating the relative contribution of different prey to the diet of an extinct predator. Several approaches have been proposed for quantifying these contributions as percentages, but the linear mixing models provide the most reliable estimates. However, these models only yield unique solutions if the number of analyzed isotopes is equal to the number of dietary sources minus one. If stable isotopes from bone collagen (δ13C, δ15N) are used, this implies that a maximum of three prey must account for the bulk of the predator's diet. Here we show that this requirement holds only for the five extant hypercarnivorous canids and probably also for the cheetah (a cursor) and the lynxes among felids, because for these species the dietary contribution of prey sources in terms of biomass is, on average, equal or higher than 55% for the first, 20% for the second and around 10% for the third. However, five to seven prey species account for the bulk of the diet of most stalking felids and also for those omnivorous canids that are not pack hunters. The conclusion is that the linear mixing models will only provide well-defined solutions using two isotopic ratios in the dietary reconstruction of extinct hypercarnivorous canids, which tend to rely very heavily on only three prey, and probably also in extremely specialized felids with cursorial adaptations or living in temperate habitats.