Accurate analyses of the diets of predators are key to understand trophic interactions and defining conservation strategies. Diets are commonly assessed through analysis of non-invasively collected scats, and the use of faecal DNA (fDNA) analysis can reduce the species misidentifications that could lead to biased ecological inference. We review the scientific literature since publication of the first paper on amplifying fDNA, in order to assess trends in the use of genetic non-invasive sampling (gNIS) for predator species identification in scat-based diet studies of North American and European terrestrial mammalian carnivores (Carnivora). We quantify error rates in morphology-based predator species identification. We then provide an overview of how applying gNIS would improve research on trophic interactions and other areas of carnivore ecology. We found that carnivore species identity was verified by using gNIS in only 8% of 400 studies of carnivore diets based on scats. The median percentage of false positives (i.e. samples wrongly identified as belonging to the target species) in morphology-based studies was 18%, and was consistent regardless of species’ body size. We did not find an increasing trend in the use of gNIS over time, despite the existing technical capability to identify almost all carnivore species. New directions for fDNA studies include employing high-throughput sequencing (HTS) and DNA metabarcoding to identify the predator species, the individual predator, the entire assemblage of consumed items, and the microbiome of the predator and pathogens. We conclude that HTS protocols and metagenomic approaches hold great promise for elevating gNIS as a fundamental cornerstone for future research in ecology and conservation biology of mammals. © 2018 The Authors. Mammal Review published by Mammal Society and John Wiley & Sons Ltd.