The molecular basis of the pathogenesis of the opportunistic invasive infections caused by isolates of the Gemella genus remains largely unknown. Moreover, inconsistencies in the current species assignation were detected after genome-level comparison of 16 public Gemella isolates. A literature search detected that, between the two most pathogenic species, Gemella morbillorum causes about twice the number of cases compared to Gemella haemolysans. These two species shared their mean diseasessepsis and endocarditis-but differed in causing other syndromes. A number of well-known virulence factors were harbored by all species, such as a manganese transport/adhesin sharing 83% identity from oral endocarditis-causing streptococci. Likewise, all Gemellae carried the genes required for incorporating phosphorylcholine into their cell walls and encoded some choline-binding proteins. In contrast, other proteins were species-specific, which may justify the known epidemiological differences. G. haemolysans, but not G. morbillorum, harbor a gene cluster potentially encoding a polysaccharidic capsule. Species-specific surface determinants also included Rib and MucBP repeats, hemoglobin-binding NEAT domains, peptidases of C5a complement factor and domains that recognize extracellular matrix molecules exposed in damaged heart valves, such as collagen and fibronectin. Surface virulence determinants were associated with several taxonomically dispersed opportunistic genera of the oral microbiota, such as Granulicatella, Parvimonas, and Streptococcus, suggesting the existence of a horizontally transferrable gene reservoir in the oral environment, likely facilitated by close proximity in biofilms and ultimately linked to endocarditis. The identification of the Gemella virulence pool should be implemented in whole genomebased protocols to rationally predict the pathogenic potential in ongoing clinical infections caused by these poorly known bacterial pathogens.