A study on histological findings in 1047 patients who underwent total axillary lymph node dissection (topographical levels I-III), was conducted. A mean of 20.3 lymph nodes (LN) was assessed; LN metastases were detected in 492 cases. Retrospective analysis showed that, by mere dissection of LNs level I, 10 of the 492 nodal positive cases (2.0%) would have been classified false negative. Additionally, according to the number of positive LNs detected in stage I, 85 of the 492 patients (17%) would have been assigned to a group with a better prognosis (prognostic groups 1-3, 4-9, greater than or equal to 10 pos. LNs). By mere dissection of LNs levels I and II, only 1 of 492 (0.2%) patients would have been categorized false negative and only 24 patients (5%) would have been included in a group with a better prognosis, on the basis of the number of positive LNs dissected. Thus dissection of LNs stages I and II provides sufficient prognostic information. Since, however, in 145 nodal positive cases (14% of the entire collective e.g. 29.5% of all patients with positive lymph nodes), metastases were detected in level III, these LNs should be removed as well, in order to prevent axillary recurrences. One could only dispense with dissection of level III, if LN metastases levels I and II could be excluded during the operation.