Colorectal surgery for cancer is associated with a high rate of surgical complications, including anastomotic leakage. The ability to predict the risk of leakage early enough seems to be of high value, since it would facilitate the design of personalized treatment and duration of hospitalization. Although different studies present the neutrophil-to-lymphocyte ratio [NLR] as having a strong predictive value, there is a discrepancy with respect to which postoperative day is the most reliable. We evaluated a series of NLR values, from the day before surgery up to the POD7, in a cohort of 245 colorectal surgery patients in order to clarify the best predictable score for the identification of the risk of anastomotic leakage. There were 28 patients with leaks. ROC curve analysis of NLR on POD1 indicates that a cut-off point ≥ 7.4 exerts a negative prediction for leakage (AUC 0.881, sensitivity 68.7%, specificity 96.4%, PPV 28.4%, and NPV of 99.3%), thus excluding 150 patients from the risk of leakage. Furthermore, the ROC curve analysis of NLR on POD4 indicates that a cut-off point ≥ 6.5 gives a positive prediction of leakage (AUC 0.698, sensitivity 82.1%, specificity 51.6%, PPV 17.6%, and NPV of 95.6%), thus indicating 52 patients as being at high risk of leakage. Finally, NLR failed to identify five leaks out of twenty-eight. These results strongly indicate the ability of NLR on POD1 to predict patients at low risk of developing a leak and then on POD4 to predict the high-risk patients. This makes our study particularly innovative, in that it enables doctors to concentrate on potential high-risk patients from POD1.