Neoadjuvant chemoradiotherapy (nCRT) is the standard treatment for locally advanced rectal cancer (LARC). Pathological complete regression is closely linked to disease outcomes. However, biomarkers predicting nCRT response and patient survival are lacking for LARC. In the present study, the clinical characteristics and follow-up information of 228 patients with LARC were retrospectively collected. Immunohistochemistry (IHC), reverse transcription-quantitative PCR (RT-qPCR), Kaplan-Meier and multivariate analyses were used to evaluate the expression and predict the role of CKLF-like MARVEL transmembrane domain member 4 (CMTM4) in LARC. Additionally, lentiviral short hairpin (sh)RNA was used to interfere with CMTM4 expression. The phenotype of CMTM4-knockdown LoVo cells was determined by colony formation, migration and invasion assays under irradiation (IR) treatment. RNA-sequencing (RNA-seq) analysis was also used to explore the CMTM4-regulated genes in LoVo-shCMTM4 cells compared with control cells. RT-qPCR was then used to confirm the expression of these CMTM4-regulated genes. CMTM4 expression in pre-nCRT tissues indicated an unfavorable response and a short disease-free survival (DFS) with LARC. The expression of CMTM4 significantly increased following nCRT treatment. Additionally, CMTM4 knockdown increased the proliferation, migration and invasion of colon cancer cells; however, IR disrupted the cell migration and invasion induced by CMTM4 knockdown. RNA-seq analysis, the Tumor Immune Estimation Resource database and RT-qPCR indicated that CMTM4 was involved in different signaling pathways and regulated immune-related genes such as cluster of differentiation 66b, chemokine (CXC motif) ligand 8 (CXCL8) and programmed cell death 1. Furthermore, CXCL8 expression was found to be negatively associated with CMTM4 expression in patients with LARC by IHC and RT-qPCR. CXCL8 expression on invasion margin regions in post-operative tissues was also an inferior predictor of DFS in patients with LARC. In conclusion, CMTM4 may predict the nCRT response and outcomes in patients with LARC.