Vaccination against avian influenza (AI) virus, a powerful tool for control of the disease, may result in issues related to surveillance programs and international trade of poultry and poultry products. The use of AI vaccination in poultry would have greater worldwide acceptance if a reliable test were available that clearly discriminated between naturally infected and vaccinated-only animals (DIVA). Because the nonstructural protein (NS1) is expressed in influenza virus-infected cells, and it is not packaged in the virion, it is an attractive candidate for a DIVA differential diagnostic test. The aim of this work was to determine the onset of the antibody response to the NS1 protein in chickens infected with low pathogenic avian influenza (LPAI) virus, and to evaluate the diagnostic potential of a baculovirus-expressed purified NS1 protein in an indirect ELISA-based DIVA strategy. An antibody response against NS1 was first detected 3 wk after infection, but the antibody levels were decreasing rapidly by 5 wk after infection. However, most chickens did not have detectable antibodies in spite of high hemagglutination inhibition (HI) antibody titers in one group. In birds vaccinated with inactivated oil-emulsion vaccines, antibodies against NS1 were not detected before virulent challenge, and only a small percentage of birds seroconverted after homologous LPAI virus challenge. Vaccinated birds challenged with highly pathogenic AI showed a higher NS1 antibody response, but at most only 40% of birds seroconverted against NS1 protein by 3 wk after challenge. Because of the variability of seroconversion and the duration of the antibody response in chickens, the NS1 protein DIVA strategy did not perform as well as expected, and if this strategy were to be used, it would require sampling a higher number of birds to compensate for the lower seroconversion rate.