Sepsis commonly occurs in patients with serious infections. It severely threatens the health of patients and has very high mortality rates. Urosepsis is a type of sepsis in which the serious infection originates from the urinary system. Early diagnosis of the occurrence and severity of urogenital sepsis is crucial for improving patient prognosis. Long noncoding RNAs (LncRNAs) play important roles in the occurrence of a number of diseases, including sepsis, and can be potential biomarkers that predict disease development. The present study aimed to discover potential LncRNAs that can predict the occurrence of urosepsis. RNA-sequence data from patients with sepsis from the GEO database was analyzed and LncRNAs associated with sepsis were identified. The expression of LncRNAs associated with sepsis was tested in clinical urosepsis samples. Finally, the value of these LncRNAs in predicting urosepsis was verified using clinical samples. From the GEO database a total of nine LncRNAs (MALAT1, NEAT1, RMRP, LncIRX5, LINC01742, DSCR4, C22ORF34, LINC00381, and LINC01102) were identified that had expression changes corresponding with the occurrence of sepsis. Specifically, MALAT1, NEAT1 and DSCR4 revealed differential expression in patients with urosepsis. Moreover, MALAT1, and DSCR4 were shown to be significant risk indicators for urosepsis, and NEAT1 was shown to reflect disease severity. Therefore, the present study indicated that the LncRNAs, MALAT1, NEAT1 and DSCR4 can reflect the occurrence and severity of urosepsis and may act as potential biomarkers.