Since November 2008, we have conducted active avian influenza surveillance in Bangladesh. Clades 2.2.2, 2.3.4.2, and 2.3.2.1a of highly pathogenic avian influenza H5N1 viruses have all been identified in Bangladeshi live poultry markets (LPMs), although, since the end of 2014, H5N1 viruses have been exclusively from clade 2.3.2.1a. In June 2015, a new reassortant H5N1 virus (H5N1-R1) from clade 2.3.2.1a was identified, containing haemagglutinin, neuraminidase, and matrix genes of H5N1 viruses circulating in Bangladesh since 2011, plus five other genes of Eurasian-lineage low pathogenic avian influenza A (LPAI) viruses. Here we report the status of circulating avian influenza A viruses in Bangladeshi LPMs from March 2016 to January 2018. Until April 2017, H5N1 viruses exclusively belonged to H5N1-R1 clade 2.3.2.1a. However, in May 2017, we identified another reassortant H5N1 (H5N1-R2), also of clade 2.3.2.1a, wherein the PA gene segment of H5N1-R1 was replaced by that of another Eurasian-lineage LPAI virus related to A/duck/Bangladesh/30828/2016 (H3N8), detected in Bangladeshi LPM in September 2016. Currently, both reassortant H5N1-R1 and H5N1-R2 co-circulate in Bangladeshi LPMs. Furthermore, some LPAI viruses isolated from LPMs during 2016-2017 were closely related to those from ducks in free-range farms and wild birds in Tanguar haor, a wetland region of Bangladesh where ducks have frequent contact with migratory birds. These data support a hypothesis where Tanguar haor-like ecosystems provide a mechanism for movement of LPAI viruses to LPMs where reassortment with poultry viruses occurs adding to the diversity of viruses at this human-animal interface.