Avian influenza (AI) is endemic and frequently causes seasonal outbreaks in winter in Bangladesh due to high pathogenic avian influenza (HPAI) H5N1 and low pathogenic avian influenza (LPAI) H9N2. Among avian influenza A viruses (AIV), H5, H7, and H9 subtypes have the most zoonotic potential. Captive birds in zoos and safari parks are used for educational, recreational, breeding, and conservational purposes in Bangladesh. To screen for AIV in captive birds to assess potential public health threats, we conducted a cross-sectional study in two safari parks and one zoo in Bangladesh for four months, from November to December 2013 and from January to February 2014. We collected blood samples, oropharyngeal, and cloacal swabs from 228 birds. We tested serum samples for AIV antibodies using competitive enzyme-linked immunosorbent assay (c-ELISA) and AIV sero-subtype H5, H7, and H9 using hemagglutination inhibition (HI) test. Swab samples were tested for the presence of avian influenza viral RNA using real-time reverse transcription-polymerase chain reaction (rRT-PCR). Across all the samples, AIV antibody prevalence was 9.7% (95% CI: 6.1-14.2, n = 228) and AIV HA subtype H5, H7 and H9 sero-prevalence was 0% (95% CI: 0-1.6, n = 228), 0% (95% CI: 0-1.6, n = 228) and 6.6% (95% CI: 3.72-10.6, n = 228), respectively. No AI viral RNA (M-gene) was detected in any swab sample (0%, 95% CI: 0-1.6, n = 228). Birds in the Safari park at Cox´s Bazar had a higher prevalence in both AIV antibody prevalence (13.5%) and AIV H9 sero-prevalence (9.6%) than any of the other sites, although the difference was not statistically significant. Among eight species of birds, Emu (Dromaius novaehollandiae) had the highest sero-positivity for both AIV antibody prevalence (26.1%) and AIV H9 prevalence (17.4%) followed by Golden pheasant (Chrysolophus pictus) with AIV antibody prevalence of 18.2% and AIV H9 prevalence of 11.4%. Our results highlight the presence of AI antibodies indicating low pathogenic AIV mingling in captive birds in zoos and safari parks in Bangladesh. Continuous programmed surveillance is therefore recommended to help better understand the diversity of AIVs and provide a clear picture of AI in captive wild birds, enabling interventions to reduce the risk of AIV transmission to humans.