A flow-through system (FT) and a recycling aquaculture system (RS), both used to culture the Pacific oyster Crassostrea gigas larvae, were compared using three concentrations of larvae, 50, 150, and 300 larvae mL-1. The physic-chemical factors and performances of larvae were measured. Given the increasing biomass load in rearing, the formation of biofilm on the tank wall was examined. The biofilm was observed by scanning electron microscopy, and bacterial assemblages were investigated by pyrosequencing the V1–V3 region of 16S rRNA gene. The best growth (13.7 µm day-1) with high survival (87%) occurred in FT at 50 larvae mL-1, while equivalent performances (growth rate around 11 µm day-1 and survival rate around 74%) were observed in RS at 50 larvae mL-1 and in both systems at 150 larvae mL-1. At 300 larvae mL-1, the larval rearing became unstable with the occurrence of mortality. Observations using scanning electron micrograph showed the formation of the biofilm over time along with the increase in its thickness and extent with larval density. The pioneer bacteria were coccobacilli, followed by filamentous bacteria. With their exudates, the latter constituted abundant braids at the end of rearing at high larval concentrations. The first colonizers were mainly affiliated to the Rhodobacteraceae (a-Proteobacteria). The filamentous bacteria corresponded to Saprospiraceae (Bacteroidetes) and Anaerolinea (Chloroflexi). Several other minor groups including Actinobacteria, Planctomyctes, d-, ?-Proteobacteria, and Flavobacteriales as well as algal cells distributed to larvae also composed this biofilm. The proportion of these different groups significantly differed between FT and RS. No pathogenic bacteria, notably Vibrio spp., were detected. The filamentous bacteria, some of which may be bactericidal and algacidal, were suspected of inducing mortality, mainly in FT at 300 larvae mL-1 where they were most abundant. These preliminary findings show the importance of the bacterial biomass constituting the biofilm in larval rearing, as it may be a source of variability in rearing. ntain pathogenic bacteria. The presence of this microbiota might