Bloodstream infections remain prevalent in intensive care units, leading to a public health challenge worldwide. Routine diagnosis is mainly based on blood culture, but the technique is limited by its time-consuming process and relatively low sensitivity. Emerging molecular diagnostic tools, such as 16S metagenomics, have been developed for detecting bacteria in the blood samples of septic patients. Using a collection of 168 blood samples from 96 septic patients, 16S metagenomics method followed by bioinformatics were applied to study bacterial alterations during the pathogenesis of sepsis. Significant taxonomic variations were found between the two survival groups at different therapeutic time points through sequential 16S metagenomics research. The results on the third day during the treatment course were notably distinct among the studied groups. 16S metagenomics approach can bring novel genetic insight about microbiological fluctuations during septic progression, which may be utilized as a complementary prognostic application. Further etiologic and pathophysiologic explorations are needed to fully explain the linkage between clinical outcomes and genetic changes.