Background The lack of effective prognostic models for colorectal cancer (CRC) makes it crucial to establish a reliable prognostic model for colorectal cancer. The aim of this study was to develop a reliable prognostic model in CRC. Using CRC samples from the TCGA database, a multi-gene combination model was screened for prognosis in CRC patients and the prognostic power of the multi-gene combination model was validated. We screened a multi-gene combination model for risk stratification and prognosis of CRC patients, which is important for individualised clinical treatment and long-term management of colorectal cancer patients. Results To solve these questions, we used data from colorectal cancer patients in the TCGA database to uncover differential genes, screened variables using one-way COX analysis and LASSO regression analysis, and then screened for the polygenic combination most associated with colorectal cancer and assessed its significance on survival events using multi-factor COX analysis and ROC curves to assess polygenic combinations. Finally, using gene enrichment analysis, the polygenic combinations were found to be associated with tRNA biosynthesis, homologous recombination, nucleotide excision repair, ribosome biogenesis in eukaryotes and DNA replication. Conclusions The multi-gene combination (FCGBP, GSTM1, IGLV4-69, IGSF9 and PCOLCE2) model screened in this study is capable of prognosticating colorectal cancer, and the findings of this study could shed new light on the prognosis of colorectal cancer in the clinical setting.