The prognostic role of the neutrophil-to-lymphocyte ratio (NLR) has been reported in colorectal cancer (CRC); however, its variation and corresponding predicative value in patients undergoing resection remain largely unknown. In the present study, data from 146 patients with CRC were retrospectively collected, optimal cutoff points for preoperative and postoperative low and high NLRs were set, and ΔNLR was calculated. Subsequently, patients were classified into low-low, low-high, high-low and high-high subgroups based on the cutoff points, and their progression-free survival (PFS) was determined. A Cox proportional hazard model was applied to calculate the prognostic value of all factors. The results demonstrated that both preoperative and postoperative NLRs (pre-NLR and post-NLR) but not ΔNLR could predict PFS with optimal cutoff points of 2.39 and 2.96, respectively. For predicting PFS, the pre-NLR had a sensitivity and specificity of 48.80 and 79.50%, respectively, and the post-NLR had a sensitivity and specificity of 63.20 and 56.20%, respectively. Significant differences were identified between low and high pre-NLRs in terms of histological grade (P<0.01) and tumor diameter (P<0.01); however, such differences were only found in terms of age (P<0.01) for low and high post-NLRs. The PFS of patients in the low-low, low-high, high-low and high-high subgroups was 50.30±21.36, 43.67±22.78, 31.06±25.56 and 29.87±24.13 months, respectively, and patients in the high-high subgroup had the worst PFS (P<0.01). Preoperative CEA level, invasive depth, node involvement, distant metastasis and preoperative NLR were independent prognostic factors. In conclusion, a persistently high NLR for patients with CRC undergoing resection was associated with poor prognosis.