Live animal markets are known hotspots of zoonotic disease emergence. To mitigate those risks, we need to understand how networks shaped by trading practices influence disease spread. Yet, those practices are rarely recorded in high-risk settings. Through a large cross-sectional study, we assessed the potential impact of live poultry trading networks' structures on avian influenza transmission dynamics in Bangladesh. Networks promoted mixing between chickens sourced from different farming systems and geographical locations, fostering co-circulation of viral strains of diverse origins in markets. Viral transmission models suggested that the observed rise in viral prevalence from farms to markets was unlikely explained by intra-market transmission alone, but substantially influenced by transmission occurring in upstream network nodes. Disease control interventions should therefore alter the entire network structures. However, as networks differed between chicken types and city supplied, standardised interventions are unlikely to be effective, and should be tailored to local structural characteristics.