Bio-slow　sand　filters　(BSSFs)　are　widely　employed　in　rural　drinking　water　treatment　ascribing　to　its　economic　applicability,　easy　operation　and　high　efficiency.　This　study　investigated　the　influence　of　particle　size,　filter　thickness,　water　quality　and　filtration　rate　on　decontamination　performance,　bacterial　community　and　metabolic　function　of　BSSFs.　The　effluent　ammonia　nitrogen　during　stable　operation　period　basically　met　the　standards　for　drinking　water　quality　in　China　(0.5　mg/L).　The　average　removal　efficiencies　of　fine　sand　were　about　4　%　higher　than　those　of　coarse　sand,　and　thick　filter　layer　showed　more　stable　performance.　Schmutzdecke　removed　most　turbidity　and　organic　pollutants,　playing　an　overwhelming　role　in　water　purification.　Filtration　rate　was　a　key　factor　in　shaping　bacterial　community　structure.　Proteobacteria　and　Cyanobacteria　were　dominant　phyla,　whose　relative　abundance　significantly　decreased　and　increased　as　filtration　rate　increasing,　respectively.　Co-occurrence　patterns　were　dominant　within　bacterial　communities.　Function　bacteria　(e.g.　Hyphomicrobium　and　Methylophilus)　and　rare　genera　(Curvibacter　and　Simplicispira)　were　identified　as　hub　genera　in　the　networks.　The　bacterial　communities　exhibited　a　metabolic　versatility.　Several　pathways　for　secondary　metabolism　significantly　shifted　under　different　conditions,　such　as　xenobiotics　biodegradation　and　metabolism.　Moreover,　filtration　rate　and　dominant　genera　greatly　affected　pollutant　removal　performance.　Many　species　of　Proteobacteria　and　Bacteroidota　were　closely　related　to　carbohydrate　metabolism,　energy　metabolism　and　nucleotide　metabolism,　and　metabolic　functions　corresponded　more　cohesively　to　conditional　abundant　taxa,　conditional　rare　or　abundant　taxa,　conditional　rare　taxa.　This　study　highlights　the　adaptability　and　bacterial　community　characteristics　of　BSSFs,　and　provides　references　for　their　practical　application.