Rationale: The host immune determinants that distinguish protective from life-threatening responses to influenza are poorly understood. Identifying drivers of immunopathology in the human lung is critical for developing potential therapies.

Objectives: To define the cellular and molecular immune landscape of the lung in mild versus severe influenza and to identify key cellular states and pathways associated with disease severity.

Methods: We generated a large-scale single-cell atlas by sequencing over 520,000 cells from the bronchoalveolar lavage fluid of 88 non-immunocompromised adult individuals with mild or severe influenza A and healthy controls. Key findings were validated by flow cytometry and protein quantification, and machine-learning models were used to identify predictive signatures.

Main results: Severe influenza was characterized by profound pulmonary lymphopenia and a massive influx of functionally dysregulated neutrophils. The infiltrating neutrophils were primed for extracellular trap formation, driving a cytokine storm via the S100A8/A9/A12-TLR4 and CXCL8-CXCR1/2 axes. This pathology coincided with the depletion and functional impairment of resident alveolar macrophages and an expansion of pro-inflammatory, monocyte-derived macrophages that amplified neutrophil recruitment. Lymphopenia in severe disease arose from synergistic cell-death programs, while remaining lymphocytes exhibited a dysfunctional state of concurrent exhaustion and hyper-cytotoxicity. Mild influenza featured a coordinated adaptive immune response, distinguished by an enrichment of T follicular helper cells and plasma cells. Machine-earning models identified robust cellular and transcriptional signatures predictive of disease severity.

Conclusions: Our atlas defines the divergent immune trajectories in influenza, revealing specific cellular states and pathways that drive immunopathology and provide novel targets for host-directed therapies.