Metagenomic next-generation sequencing (mNGS), which provides untargeted and unbiased pathogens detection, has been extensively applied to improve diagnosis of pulmonary infection. This study aimed to compare the clinical performance between mNGS and targeted NGS (tNGS) for microbial detection and identification in bronchoalveolar lavage fluid (BALF) from kidney transplantation recipients (KTRs). BALF samples with microbiological results from mNGS and conventional microbiological test (CMT) were included. For tNGS, samples were extracted, amplified by polymerase chain reaction with pathogen-specific primers, and sequenced on an Illumina Nextseq. A total of 99 BALF from 99 KTRs, among which 93 were diagnosed as pulmonary infection, were analyzed. Compared with CMT, both mNGS and tNGS showed higher positive rate and sensitivity (p<0.001) for overall, bacterial and fungal detection. Although the positive rate for mNGS and tNGS was comparable, mNGS significantly outperformed tNGS in sensitivity (100% vs. 93.55%, p<0.05), particularly for bacteria and virus (p<0.001). Moreover, the true positive rate for detected microbes of mNGS was superior over that of tNGS (73.97% vs. 63.15%, p<0.05), and the difference was also significant when specific for bacteria (94.59% vs. 64.81%, p<0.001) and fungi (93.85% vs. 72.58%, p<0.01). Additionally, we found that, unlike most microbes such as SARS-CoV-2, Aspergillus, and EBV, which were predominantly detected from recipients who underwent surgery over 3 years, Torque teno virus (TTV) were principally detected from recipients within 1-year post-transplant, and as post-transplantation time increased, the percentage of TTV positivity declined. Although tNGS was inferior to mNGS owing to lower sensitivity and true positive rate in identifying respiratory pathogens among KTRs, both considerably outperformed CMT.