The management of stage II colorectal cancer is still difficult. We aimed to construct a new immune cellassociated signature for prognostic evaluation and guiding chemotherapy in stage II colorectal cancer. We used the "Cell Type Identification by Estimating Relative Subsets of RNA Transcripts" (CIBERSORT) method to estimate the fraction of 22 immune cells by analyzing bulk tumor transcriptomes and a LASSO Cox regression model to select the prognostic immune cells. A 12-immune cell prognostic classifier, ISCRC, was built, which could successfully discriminate the high-risk patients in the training cohort (GSE39582: HR = 3.16, 95% CI: 1.85-5.40, P < 0.0001) and another independent cohorts (GSE14333: HR = 3.47, 95% CI: 1.18-10.15, P =0.0167). The receiver operating characteristic analysis revealed that the AUC of the ISCRC model was significantly greater than that of oncotypeDX model (0.7111 versus 0.5647, p=0.0152). We introduced the propensity score matching analysis to eliminate the selection bias; survival analysis showed relatively poor prognosis after chemotherapy in stage II CRC patients. Furthermore, a nomogram was built for clinicians and did well in the calibration plots. In conclusion, this immune cell-based signature could improve prognostic prediction and may help guide chemotherapy in stage II colorectal cancer patients.