Aberrant methylation is strongly associated with development of cancer, but limited data exist on correlation between methylation and regional lymph node metastasis (RLNM). The aim of this research was to study using of methylation levels of WIF1, RASSF1A, CDO1 and MEST aberrant methylated genes in a primary breast cancer for prediction of regional lymph node metastases. We used MS-HRM (Methylation Sensitive High Resolution Melting) to assess methylation levels. The results were confirmed by pyrosequencing. The study included 66 women with LumA and 46 women with HER2- (LumB-), 22 and 26 of them had metastasis in at least one lymph node respectively. It was found that methylation levels between LumA and LumB subtypes differed significantly in genes: WIF1 (p<0.001), CDO1 (p=0.002) and MEST (p=0.033). In the Lum A subtype statistically significant differences in level of methylation of WIF1 gene between patients with metastases in RLNM and patients without metastases were found (p=0.03). Analysis of tumors longer than 2 cm in the LumA subtype, revealed an increase of statistical significance of WIF1 gene - p=0.009 (AUC (95%CI) = 0.76 (0.59-0.93)). In LumB- subtype RASSF1A, CDO1 and MEST had statistically significant differences in methylation level between groups (p=0.03, p=0.048 and p=0.045 respectively). ROC analysis showed that combining of three genes by logistic regression, AUC (95%CI) was 0.74 (0.6-0.88). Analysis of tumors longer than 2 cm, did not increase statistical significance for these genes (p=0.046; p=0.089 and p=0.076, respectively). Thus, the study of methylation in primary tumors may be useful for prediction of lymph node metastasis, as well as for better understanding of biological process inside breast cancer.