H9N2 is currently the second most common avian influenza A virus subtype infecting humans. Monitoring viral phenotypic and genotypic adaptation to humans is crucial for risk assessment. Here, we compared the replication of an H9N2 human isolate collected in 2024 (A/HK/2346/2024) to a historical human isolate collected in 1999 (A/HK/1073/1999). In Madin Darby canine kidney (MDCK) cells, A/HK/2346/2024 and A/HK/1073/1999 replicated to 8 log10 and 5 log10 plaque-forming units (PFU) per ml, respectively. In both human nasal and lung organoids, A/HK/2346/2024 replicated to 6 log10 PFU/ml, but A/HK/1073/1999 failed to replicate in either organoid. The infection rates of both ciliated and non-ciliated cells were higher for A/HK/2346/2024 than A/HK/1073/1999, and the ratios of infected 2,6/2,3 cells were also higher for A/HK/2346/2024. Apart from the mammalian adaptive substitutions that were present in the nasopharyngeal specimen collected on day 1 post-symptom onset (pso) (HA-D183N/D190 T/Q192R/Q226L; NA-del62-64; PB2-A588 V/K702R; PB1-I368 V; PA-K356R/S409N; M1-R95 K), the mammalian adaptive substitution PB2-D253N emerged de novo on day 7 pso. Analysis of all human (n = 96) and avian influenza (n = 14762) H9N2 deposited at GISAID showed the dominance of several human-adaptation substitutions in H9N2 strains collected from humans in recent years. In summary, we demonstrated that a recent H9N2 virus is more adapted to humans, and is able to replicate to high titers in both upper and lower human respiratory tract which may confer higher person-to-person transmissibility and virulence. Our study underscores the importance of human organoid-based phenotypic monitoring and inter/intrahost genotypic monitoring for assessing the zoonotic risk of avian influenza viruses.