Abstract: Objective To develop a prognosis model for diffuse large B-cell lymphoma （DLBCL） based on the perspective of molecular biology. Methods The DLBCL gene-expression data and clinical information were downloaded from the GEO and TCGA databases. The Gene prediction model was constructed by using Lasso regression and multivariable Cox regression，and the gene model was combined with other clinical prognostic factors to construct the nomogram model. Results The prediction model was established by six genes （LNPEP、SNX20、GTPBP10、CALR、BDH1、C5orf30） which could divide patients in each cohort into high-risk group and low-risk group. The AUC of 3-year overall survival rate in GSE10846，GSE32918 and NCICCR were 0.722，0.758 and 0.693，respectively. Based on this gene model and the clinical factors such as age，subtype，therapeutic schedule，ECOG，stage，and number of junction external locus，we concluded that the AUC of 3-year overall survival rate in GSE10846 was 0.796，the calibration plot showed good consistency. The GO and KEGG enrichment analysis showed that the genes of this model were mainly related to the biological functions and pathways such as DNA replication，repair，protein processing，cell cycle and viral carcinogenesis. Conclusions A gene prediction model is constructed，which can be used to predict the survival of DLBCL patients. When combined with clinical factors，the model can be better used to evaluate the prognosis of patients.

Key words: Diffuse large B-cell lymphoma, Gene based prognostic model, Risk score