BACKGROUND: The emergence of highly pathogenic avian influenza H5N1 viruses has raised concerns about their pandemic potential. Vaccination is the most effective way of preventing influenza. In this study, we investigated the safety and immunogenicity of an avian H5N2 live attenuated influenza vaccine (LAIV H5N2) in healthy Thai adults and its priming immune responses with an H5N1 inactivated vaccine boost. METHODS: This study was done at the Vaccine Trial Centre at Mahidol University, Bangkok, Thailand and was divided into two parts. Part 1 consisted of a randomised, double-blind, placebo-controlled trial done over 18 months. We randomly assigned (2:1) healthy Thai adults aged 18-49 years with a computer generated randomisation sequence (blocks of six) to receive either two intranasal doses (0·25 mL per nostril) of LAIV H5N2 (101 participants) or placebo (51 participants) 21 days apart. For part 2, an open-label trial was done in which previously vaccinated participants (40 from LAIV H5N2 group and 20 placebo) were given one intramuscular dose (0·5 mL) of H5N1 booster vaccine. Participants, investigators, and site-study workers were blinded from randomisation. Immune responses after subsequent immunisation were evaluated using haemagglutination-inhibition and microneutralisation assays and circulating follicular T-helper cells and plasmablast cells were measured in serum and whole blood. The trials are registered with ClinicalTrials.gov, numbers NCT01841918 and NCT02229357. FINDINGS: Between Feb 4, 2013, and Feb 28, 2013, 256 individuals were screened, of whom 152 participants were enrolled in part 1 of this study. LAIV H5N2 vaccine was well tolerated. Viral shedding was detected in only six (6%) of 101 participants in the vaccine group 1 day after the first vaccination and in and two (2%) of 98 participants in the group after the second vaccination. There was no serious adverse event in both groups. 51 (50%) of 101 participants in the vaccine group and 28 (55%) of 51 in the placebo group reported at least one adverse event. 80 (84%) of 95 events in the vaccine group and 32 (78%) of 43 events in the placebo groups were reportedly suspected adverse events, probably related to the vaccine; however, most were mild in nature. After two doses of vaccine, 13 (13%) of 100 participants in the vaccine group had an increase in haemagglutination-inhibition titre of more than four-fold and four (4%) of 100 vaccinees developed a rise in neutralisng antibody titre of more than four-fold. 1 year later, after a booster with an inactivated H5N1 vaccine (part 2), 39 (98%) of 40 participants who had previously been vaccinated with LAIV H5N2 had an increase in haemagglutination-inhibition titre of greater than four-fold as early as day 7 compared with three (15%) of 20 participants in the placebo group. Peak geometric mean titre (GMT) for haemagglutination-inhibition antibodies in the previously LAIV H5N2 vaccinated group (566·89 [95% CI 436·97-735·44]) were significantly higher than among those who previously received placebo (25·49 [11·82-54·96]; p<0·0001). The peak GMT by neutralising antibody assay in the H5N2 vaccinated group (1395·85 [1040·79-1872·03]) was also significantly higher than that observed in the placebo group (17·41 [9·05-33·48]; p<0·0001). Importantly, higher cross-reactive haemagglutination-inhibition antibody titres against H5N1 (clades 1, 2.1.3.2, and 2.3.4) were detected in the LAIV H5N2 experienced group than the naive group (p<0·0001). INTERPRETATION: Our data suggest that LAIV vaccination induces long-lasting memory immune responses. The limitation of this study was that part 2 was designed as a proof-of-concept study by contrast with part 1. FUNDING: WHO.