Lymphedema (LE) is a well-known postoperative complication after axillary node dissection (ALND). Although, sentinel lymph node dissection (SLND) involves more focused surgery and less disruption of the axilla, early reports show up to 13% of patients experience some symptoms of LE. The purpose of this study was to determine predictors of arm LE in our patients under going SLND with or without an ALND. One hundred and thirty-seven breast cancer patients were treated at a comprehensive cancer center. Prospective measurement of arm volume was carried every 6 months from date of diagnosis. This data base was retrospectively reviewed for tumor stage, treatment, and subjective complaints of LE. Objective LE was defined as a change greater than 200 mL compared with the control arm. Univariate and multivariate analyses were performed. Arm volume changes were measured over 24 months (median follow-up 20 months) in 137 women: 82 stage I, 48 stage II, and 5 stage III; median age 56 years. Breast-conserving surgery was performed in 133 patients. All patients underwent SLND for axillary staging and for 52 patients this was the only axillary staging procedure. All node-positive patients (31) and 54 node-negative patients under went an immediate completion ALND, the latter as part of a study protocol. At 24 months, 16 (11.6%) patients were found to have objective LE (>200 mL increase). Patient age, tumor size, number of nodes harvested, or adjuvant chemotherapy was not found to be predictive of LE by univariate analysis. The risk of developing postoperative LE was primarily and significantly related to the patients' BMI (p = 0.003). Multivariate analysis revealed patients with a BMI >30 (obese) had an odds ratio of 2.93 (95% CI 1.03-8.31) compared with those with a BMI of <25 of having LE. Symptomatic LE (SLE), as defined by patient complaints was recorded in six of the above 16 patients, no SLE was recorded in patients without objective signs of edema. Univariate subgroup analysis compared the symptomatic to the nonsymptomatic patients and revealed the median number of nodes removed was higher in the symptomatic patients (17 verses 9, p = 0.045); however, these patients had a lower BMI (p = 0.0012). The mean change in arm volume was not significantly different between the groups. SLE occurs in one third of patients with objective arm swelling and most likely is multi-factorial in etiology. Although patients undergoing SLN were recorded as having objective LE, none reported SLE. The development of LE within 2 years of surgery is associated with the patient's BMI and this should be considered in preoperative counseling.