Sepsis is a life-threatening infection demanding accurate and fast diagnosis of pathogen for right antibiotics. In current practice, culture-based methods are still the standard approach of etiological diagnosis of sepsis. Practically, these methods have disadvantages including long turnaround time and low sensitivity. Although molecular assays, such as polymerase chain reaction (PCR)-based methods have been established for sepsis, the spectrum of the detection is limited to known pathogens listed on the panel. In the last few years, the next-generation sequencing (NGS) has been constantly improving and stepped into clinical practice, providing a powerful tool with advantages including short turnaround time, full spectrum and semiquantitative analysis. Although NGS has clear advantages compared with culture-based methods, we are facing the following challenges: few acknowledged criteria of sequencing result explanation have been established; the relationship between sequencing results and treatment effects is undetermined; and the drug resistance gene detection is usually with low sensitivity. Besides, large-scale study comparing NGS with traditional etiological diagnosis methods is missing. Thus, NGS still needs high-level evidence to support its clinical application.