Purpose Sepsis is an unusual systemic reaction to what is sometimes an otherwise ordinary infection, and it probably represents a pattern of response by the immune system to injury. However, the relationship between biomarkers and sepsis remains unclear. This study aimed to find potential molecular biomarkers, which could do some help to patients with sepsis.Methods The sepsis dataset GSE28750, GSE57065 was downloaded from the GEO database, and ten patients with or without sepsis from our hospital were admitted for RNA-seq and the differentially expressed genes (DEGs) were screened. The Metascape database was used for functional enrichment analysis and was used to found the differential gene list. Protein-protein interaction network was used and further analyzed by using Cytoscape and STRING. Logistic regression and Correlation analysis were used to find the potential molecular biomarkers.Results Taking the intersection of the three datasets yielded 287 differential genes. The enrichment results included Neutrophil degranulation, leukocyte activation, immune effectors process, positive regulation of immune response, regulation of leukocyte activation. The top 10 key genes of PPI connectivity were screened using cytoHubba plugin, which were KLRK1, KLRB1, IL7R, GZMA, CD27, PRF1, CD8A, CD2, IL2RB, and GZMB. All of the hub genes are higher expressed in health group of different databases. Logistic regression showed that IL7R, GZMA and CD8A proteins were analyzed and all of them were statistically significant. Correlation analysis showed that there was a statistically significant correlation between IL7R, GZMA and CD8A.Conclusion KLRK1, KLRB1, IL7R, GZMA, CD27, PRF1, CD8A, CD2, IL2RB, GZMB are key genes in sepsis, which associated with the development of sepsis. However, IL7R, GZMA and CD8A may serve as the attractively potential molecular biomarkers for sepsis.