Edema of the arm can be a significant complication following treatment of breast cancer. To determine the risk of arm edema and factors associated with this risk in patients treated with primary radiotherapy, we reviewed the records of 475 women with early breast cancer treated between 1968 and 1980. During this period, the use of axillary surgery prior to radiation gradually increased, and all patients received full axillary irradiation until late in the series. Based on the surgeon's report, the extent of axillary surgery was classified as either a sampling, a lower dissection, or a full dissection. Edema of the arm was scored on clinical grounds and ranged from mild hand swelling to an increased arm circumference of 8 cm. At 6 years, the actuarial risk of developing arm edema was 8% for the entire study population. This risk was 13% for 240 patients who had axillary surgery and 4% for 235 patients not undergoing axillary surgery (p = 0.006). For patients undergoing axillary surgery, the risk of arm edema was 37% with full dissection compared to 5% with sampling (p = 0.0003), and 8% with lower dissection (p = 0.03). The risk of arm edema at 6 years was 28% if more than ten nodes were removed, and 9% if one to ten nodes were removed (p = 0.03). However, the extent of axillary dissection was stronger predictor of subsequent edema than was the number of nodes obtained. The role of axillary irradiation could not be evaluated since 91% of patients received axillary irradiation. The use of chemotherapy, the site or size of the primary tumor, clinical nodal status, patient age and weight, type of suture, the use of a drain, and subsequent local or distant failure did not appear to be significant risk factors. We conclude that the combination of full dissection and full axillary irradiation results in an unacceptably high risk of arm edema.