Lymphovenous anastomoses (LVA) offer ideal physiologic treatment for lymphedema, and our experimental data support late patency. Between Jan. 1, 1982, and April 1, 1986, 18 patients underwent operation for chronic lymphedema; LVA could be performed in 14 patients (10 women and four men). Six patients had secondary lymphedema of the upper extremity. One of eight patients with lymphedema of the lower extremity had filariasis, and seven had primary lymphedema. Mean follow-up was 36.6 months (range: 5 to 57 months). Limb circumference and volume, number of postoperative episodes of cellulitis, and lymphoscintigraphy were used to assess results. Improvement occurred in three upper extremities and two lower extremities. There was no change in five extremities, and in four patients the edema progressed. One patient with primary lymphedema and four of seven patients with secondary lymphedema improved. Only one of five patients benefited from one anastomosis; however, all patients with more than two anastomoses improved. Lymphoscintigraphy was performed in 10 patients. No lymphatic channel was visualized before operation in three patients, and at operation none was found. In four other patients lymph channels localized by lymphoscintigraphy were identified during operation. Significant improvement was documented by lymphoscintigraphy in one patient after operation, and this patient had permanent improvement 30 months later. Patients with primary lymphedema had disappointing results, but four of seven patients with secondary lymphedema benefited from LVA, especially if several anastomoses could be performed. Lymphoscintigraphy appears to be a suitable method of both identifying patent lymph channels before surgery and determining function of LVA after operation. However, presently objective data to prove the clinical efficacy of this operation are lacking.