Nodal status is the most significant independent prognostic factor in breast cancer. Identification of molecular markers would allow stratification of patients who require surgical assessment of lymph nodes from the large numbers of patients for whom this surgical procedure is unnecessary, thus leading to a more accurate prognosis. However, up to now, the reported studies are preliminary and controversial, and although hundreds of markers have been assessed, few of them have been used in clinical practice for treatment or prognosis in breast cancer. The purpose of the present study was to determine whether protein phosphatase Mg2+/Mn2+ dependent 1D, ß-1,3-N-acetylglucosaminyltransferase, neural precursor cell expressed, developmentally down-regulated 9, prohibitin, phosphoinositide-3-kinase regulatory subunit 5 (PIK3R5), phosphatidylinositol-5-phosphate 4-kinase type IIα, TRF1-interacting ankyrin-related ADP-ribose polymerase 2, BCL2 associated agonist of cell death, G2 and S-phase expressed 1 and PAX interacting protein 1 genes, described as prognostic markers in breast cancer in a previous microarray study, are also predictors of lymph node involvement in breast carcinoma Reverse transcription-quantitative polymerase chain reaction analysis was performed on primary breast tumor tissues from women with negative lymph node involvement (n=27) compared with primary tumor tissues from women with positive lymph node involvement (n=23), and was also performed on primary tumors and paired lymph node metastases (n=11). For all genes analyzed, only the PIK3R5 gene exhibited differential expression in samples of primary tumors with positive lymph node involvement compared with primary tumors with negative lymph node involvement (P=0.0347). These results demonstrate that the PIK3R5 gene may be considered predictive of lymph node involvement in breast carcinoma. Although the other genes evaluated in the present study have been previously characterized to be involved with the development of distant metastases, they did not have predictive potential.