In the absence of robust active surveillance of avian influenza viruses (AIV) affecting poultry in South Asian countries, monitoring of live bird markets (LBMs) can be an alternative. In a longitudinal study of 32 LBM, five environments were sampled as follows: market floor, stall floor, slaughter area, poultry holding cage and water used for meat processing. Samples were taken monthly for 5 months, September 2013-January 2014. Incidence rates (IRs) of LBM contamination with AIV and its subtypes H5, H7 and H9 were assessed. In 10 of the LBM selected, biosecurity measures had been implemented through FAO interventions: the other 22 were non-intervened. Standard procedures were applied to detect AIV and three subtypes in pooled samples (1:5). An LBM was considered positive for AIV or a subtype if at least one of the pooled samples tested positive. The incidence rates of LBM contamination with AIV, H5, H7 and H9 were 0.194 (95% confidence interval (CI) 0.136-0.276), 0.031 (95% CI 0.013-0.075), 0 and 0.175 (95% CI 0.12-0.253) per LBM-month at risk, respectively. The log IR ratio between the FAO-intervened and non-intervened LBM for contamination with AIV was -0.329 (95% CI -1.052 to -0.394, p = .372), 0.598 (95% CI -1.593 to 2.789, p = .593) with subtype H5 and -0.500 (95% CI -1.249 to 0.248, p = .190) with subtype H9, indicating no significant difference. The results obtained suggest that both H5 and H9 were circulating in LBM in Bangladesh in the second half of 2013. The incidence of contamination with H9 was much higher than with H5.