Forty women were investigated 6 to 24 months after radical mastectomy for the presence of functioning lymphovenous communications in the arm; 20 cases were edematous and the rest were not. Ten normal volunteer women were also investigated and served as controls. Iodinated (I125) human serum albumin was injected intralymphatically in the arm, and samples of blood were taken from both basilic veins simultaneously at 5, 10, 15, 30, 45, and 60 minutes. The radioactivity of each sample was determined with a scintillation counter. Lymphangiography was performed in all cases 1 month after the isotopic studies and lymphangiograms were evaluated for the presence of lymphovenous shunts. In the volunteers and the edematous postmastectomy group, there was negligible lymphovenous shunting of labeled albumin. In the nonedematous postmastectomy group, there was an increased quantity of local lymphovenous transfer of iodinated albumin. Lymphovenous shunts were detected radiologically in two patients who did not have edema. On the basis of our observations, the incidence of postmastectomy lymphedema is predetermined and is a consequence of the lack of existence of lymphovenous communications that can allow adequate lymph flow drainage from the arm after radical dissection of axillary lymph nodes.