The digestive tracts of vertebrates are colonized by complex assemblages of micro-organisms, collectively called the gut microbiota. Recent studies have revealed important contributions of gut microbiota to vertebrate health and disease, stimulating intense interest in understanding how gut microbial communities are assembled and how they impact host fitness (Sekirov et al. 2010). Although all vertebrates harbour a gut microbiota, current information on microbiota composition and function has been derived primarily from mammals. Comparisons of different mammalian species have revealed intriguing associations between gut microbiota composition and host diet, anatomy and phylogeny (Ley et al. 2008b). However, mammals constitute <10% of all vertebrate species, and it remains unclear whether similar associations exist in more diverse and ancient vertebrate lineages such as fish. In this issue, Sullam et al. (2012) make an important contribution toward identifying factors determining gut microbiota composition in fishes. The authors conducted a detailed meta-analysis of 25 bacterial 16S rRNA gene sequence libraries derived from the intestines of different fish species. To provide a broader context for their analysis, they compared these data sets to a large collection of 16S rRNA gene sequence data sets from diverse free-living and host-associated bacterial communities. Their results suggest that variation in gut microbiota composition in fishes is strongly correlated with species habitat salinity, trophic level and possibly taxonomy. Comparison of data sets from fish intestines and other environments revealed that fish gut microbiota compositions are often similar to those of other animals and contain relatively few free-living environmental bacteria. These results suggest that the gut microbiota composition of fishes is not a simple reflection of the micro-organisms in their local habitat but may result from host-specific selective pressures within the gut (Bevins & Salzman 2011).