Arm lymphedema (ALE) was evaluated in 74 patients treated conservatively for breast cancer. ALE was defined based upon measurements performed upon 35 volunteer subjects who did not have and were never treated for breast cancer. Multiple variable statistical analysis of 74 breast cancer patients revealed that age at diagnosis was the most important factor related to the subsequent development of ALE. ALE appeared in 7 of 28 patients (25%) 60 years of age or older but in only 3 of 46 (7%) younger patients (p less than 0.02). Axillary node dissection (AND) was the only other statistically significant factor. For the younger patients, obesity and post-operative wound complications appeared to be contributing factors. For the older patients, AND technique was the only significant factor. ALE developed in only 1 of 10 (10%) of the older patients who underwent AND without splitting the pectoralis minor muscle (PMM), but in 6 of 11 (55%) who underwent AND with PMM split (p less than 0.03). Splitting the PMM during AND did not yield more lymph nodes for pathological analysis nor did it yield a higher incidence of patients with nodal metastases. Neither the use of lymph node radiation therapy fields, radiation to the full axilla, nor systemic chemotherapy was associated with ALE. We conclude that older patients are at higher risk of ALE and that this complication can possibly be reduced by not splitting the PMM during axillary node dissection.