As the trend is towards conservative surgery for the breast carcinoma, the necessity of the complete axillary dissection is being questioned much more today. In this study, we aimed to analyze the frequency of level III lymph node metastases and the contributing risk factors. Eighty-seven female, histopathologically proven breast carcinoma patients underwent modified radical mastectomy and complete axillary dissection in the Department of First General Surgery, Ankara Oncology Hospital. The level III specimen was sent separately to the pathological examination. Age, menopausal status, tumor location, histopathological type, grade, pathological T and N stage, estrogen (ER) and progesterone (PR) receptor status, multicentricity, total metastatic lymph nodes in level I+II, lymph node capsule invasion (N1b3, N2) were analyzed as the risk factors. Mean age of the 87 patients was 49±11.6 (median: 48, min.: 24, max.: 75). Most of the patients were premenopausal (52.9%, 46/87) and had their tumors located in the upper outer quadrant (64.4%, 56/87) with the histopathological type of invasive ductal carcinoma (88.5 %, 77/87). Forty-four of the cases (50.6%) were in T2 stage. Fifty-one patients (58.6%) had grade 2 tumors. The pathological workup revealed ER positivity in 52.9% patients (46/87), PR positivity in 31% patients (27/87), lymph node capsular invasion (N1b3 and N2) in 37.9% (33/87) and multicentricity in 16.1% patients (14/87). There were 7 patients (8%) in Stage I, 18 (20.4%) in Stage IIA, 30 (34.5%) in Stage IIB and 32 (36.8%) in Stage IIIA. Sixty-two out of 87 (71.2%) had axillary lymph node metastases. Lymph node metastases in level I+II were detected in 61 patients (70.1%), where level III involvement were detected in 27 patients (31%). Forty-two percent of the patients that had involved level I+II lymph nodes also had metastases in level III. Level III skip metastasis was detected in one patient (1.14%). The stage, lymph node capsular invasion and the presence of more than 2 metastatic nodes in level I+II were found to be statistically significant for level III metastases. According to our results for appropriate staging and adequate local control we recommend complete axillary dissection including level III lymph nodes except for a selected group of patients. [Turk J Cancer 2007;37(3):109-113]