In a re-analysis of the 16S rRNA sequences from the Human Microbiome Project, Ding and Schloss show that the bacterial communities from the 18 sampled body sites can be partitioned into distinct 'community types'. The authors identified between two and seven community types per body site, which, when combined with demographic and life history data, were strongly associated with several factors, including gender and level of education. Interestingly, although the compositions of the oral and gut microbiomes are different, the community types at these sites were predictive of each other. Furthermore, community types from the oral cavity were the least stable, whereas those in the gut and vagina were the most stable. Similar to the enterotype concept for the gut microbiome, the authors argue that the considerable intrapersonal and interpersonal variation of the human microbiome can be interpreted in terms of community types, which should improve our understanding of the influence of the microbiome on human health.