Objective To investigate the clinical application of bronchoalveolar lavage fluid (BALF) metagenomics next-generation sequencing (mNGS) in programmed cell death protein 1 (PD-1) immune checkpoint inhibitor. Methods Retrospectively analyzed 30 cases with advanced non-small cell lung cancer (non-small cell lung cancer, NSCLC) who had CIP due to the combination of PD-1 inhibitors and chemotherapy. BALF mNGS, BALF/sputum culture and acid-fast staining smears were tested, and white blood cell count, high-sensitivity C-reactive protein (Hs-CRP), procalcitonin (PCT), 1,3-β -D-glucan (G test), Galactomannan (GM) in the serum were examined on the same day of tracheoscopy. The mNGS-negative group was pre-diagnosed as "CIP" and treated with standard glucocorticosteroid anti-inflammatory treatment, while the mNGS-positive group was pre-diagnosed as "CIP combined with pulmonary infection" and treated with the combination of standard glucocorticosteroid and anti-infection treatment. Results Both groups of patients improved after treatment and pulmonary infection can be completely excluded in mNGS-negative cases. For CIP patients combined with pulmonary infection, detection rate by mNGS was higher than by traditional pathogenic microbiological detection methods, especially in mixed infection and opportunistic infection. Conclusions BALF mNGS detection performs better in the diagnosis of CIP. Compared with traditional pathogenic microbial detection methods, BALF mNGS provides higher detection rate and accuracy for CIP combined with pulmonary opportunistic and mixed infection, especially for fungal infection, which provides an early detection for fungal type and load analysis method.