The approach of wastewater treatment has shifted towards a holistic view in order to achieve sustainability in addition to environmental protection. The aerobic granular sludge (AGS) technology, which relies on the use of fast-settling granular biofilms called granules, is progressively becoming a new standard for high-rate biological nutrient removal (BNR) and secondary clarification in single sequencing batch reactors (SBR). The AGS intensive process has been related with definite savings in land area, construction, and operation costs. In an economical analysis, theoretical savings of CHF 0.45 per m3 (1 Swiss franc CHF ≈ 0.83 €) were computed for a Swiss wastewater treatment plant (WWTP) of 200’000 capita removing all nutrients biologically. Besides scale-up, fundamental research was required to understand and tailor the structure of underlying bacterial communities that form granules and remove nutrients. Mechanisms of bacterial selection were investigated in a systems approach to propose strategic axes for optimal management of the bacterial resource for efficient process performances. This research led to the following advances. After having designed a flexible reactor infrastructure for AGS research, a mathematical modeling methodology was developed to understand the hydraulic and biological processes involved during plug-flow transport of wastewater across the settled AGS bed during the feeding phase. BNR relies on the preferential selection of polyphosphate-accumulating organisms (PAO) by proper uncoupled supply of electron donors and electron acceptors. A bed height to diameter ratio of 6 was deduced to be optimal for full bed loading and efficient anaerobic acetate uptake under reference simulation conditions (20°C, pH 7.0). A bioinformatics methodology called PyroTRF-ID was developed for the identification of bacterial relatives involved in BNR processes by combination of terminal-restriction fragment length polymorphism (T-RFLP) and pyrosequencing data. This procedure provided high resolution of bacterial community structures and dynamics of AGS systems. Electrical conductivity and polyphosphatase assays were proposed for rapid and low-cost assessment of the fractions of active PAO and of the dephosphatating potential of activated sludge and granular sludge. Positive linear correlations were obtained between the fraction of PAO, the biomass specific rate of conductivity evolution measured in anaerobic metabolic batch tests, and the polyphosphate-hydrolyzing enzymatic activity of cell extracts. Wash-out conditions used to stimulate granule formation were shown to exert a selection pressure not only on physical properties of early-stage granules, but also on the underlying bacterial community composition. Operation with constant volumetric organic loading rates (OLR) during wash-out resulted in the formation of slow-settling fluffy granules dominated by filamentous Burkholderiales affiliates formed under low aeration (< 2 cm s-1) and high mesophilic temperature (30°C), and of fast-settling dense granules dominated by Zoogloea spp. under high aeration (4 cm s-1) and with an inoculum originating from a BNR-WWTP. These exopolysaccharide producers were considered as model organisms for granulation. However, their proliferation over PAO and nitrifiers correlated with poor BNR. PAO estab-lished at mature stage as soon as sufficient AGS was present to fully remove volatile fatty acids (VFA) during the anaerobic feeding phase. Without purge of excess sludge, glycogen-accumulating organisms (GAO) competed with PAO at a later stage of reactor operation. Granulation was shown to occur in stirred-tank SBRs operated under steady-state condi¬tions to cultivate PAO and GAO enrichments. The Accumulibacter- and Competibacter-dominated communities displayed granulation potential without involvement of Zoogloea spp. Fast-settling nuclei formed despite initial operation with a settling time as high as 60 min, and evolved towards granules after decrease of settling time to 10 min. Dynamic control of the OLR, anaerobic contact time, and sludge retention time (SRT) selected for active PAO in early-stage granules. Confocal laser scanning microscopy (CLSM) revealed that granula¬tion mechanisms depend on the predominant organisms involved. Fast-growing Zoogloea spp. formed smooth biofilm continuum matrices. The slower-growing PAO and GAO formed het¬erogeneous granular structures by proliferation in compact colonies around flocs. Fluctuations in operation variables were shown by multivariate statistics to impact on process performance and bacterial community structures in anaerobic-aerobic AGS-SBRs. The size of granules impacted on nitrification and dephosphatation by affecting oxygen mass transfer. Efficient BNR was obtained with at least 500 mgCOD L-1of acetate, 30 gCOD gP-1, and 10 gCOD gN-1. Although clades of PAO can denitrify, nitrogen removal was thus not restricted only to PAO. A broad denitrifying community was identified. The bacterial community continuum comprised two major opponent clusters, composed of members of the core microbiome of full-scale BNR-WWTPs. The first mainly comprised Accumulibacter, Nitrospira, Xanthomonadaceae, and Aminobacter affiliates selected by conditions of efficient BNR. The second mainly comprised Competibacter, Sphingobacteriales, Cytophaga, and Tetrasphaera affiliates correlating with periods of low BNR. Bacterial selection in AGS systems was mainly impacted by pH, according to multifactorial experiments. Accumulibacter and BNR were favored with alkaline pH, low mesophilic temperatures, and in the presence of propionate. Competibacter proliferated with acidic pH, temperature close to 30°C, and only acetate. Tetrasphaera spp., potential major PAO of certain full-scale plants, withstood Competibacter-selective conditions. Enhanced BNR was obtained under non-limiting conditions with food-to-microorganism ratios above 25 mgCODs gCODx-1. With fixed 2-h starvation, dephosphatation was only complete with full aeration. Since alternating aerobic-anoxic starvation conditions led to partial conversions, the length of starvation phases should be controlled in function of the redox conditions applied. Overall, optimal bacterial resource management in AGS systems requires full control of system behavior by taking advantage of the flexibility of the SBR technology. A methodology is proposed to this end, comprising milestones for wastewater characterization, SBR design, “anaerobic selector” design, control of the starvation phase length, shaving of fluctuations in operation variables, purge of excess sludge, proper selection for a stable, active and cooperative bacterial community, as well as efficient BNR. As key research perspective, investigations with real wastewater should validate the knowledge gained on bacterial behaviors under low-complexity conditions. Significant differences in predominant bacterial relatives were detected between lab-scale and full-scale BNR sludge. Metagenomics will provide a broader view on key organisms and functions of BNR and AGS microbiomes.