The more important of the several routes of dispersal in cancer of the cervix is not easy to establish, as occasional evidence of both direct and lymphatic spread is found in the grossly early clinical case. That the type and degree of dispersion are important in the selection of therapy is apparent, but the presence or absence of early extension is difficult, and frequently impossible, to predicate. Both clinical and pathological observations recognize the convenience of separating the major lymph nodes into a primary group, which includes the paracervical, ureteral, obturator, hypogastric, and external iliac nodes, and a secondary group which includes the common iliac, aortic. sacral, and inguinal nodes (Fig. 1). The three main channels draining the cervix permit a fairly constant pattern of metastatic spread, and in most instances, one or more nodes of the primary group are involved before the level of the secondary group is reached. However, rich intercommunicating channels and anomalous nodes account for the unpredictable sites and routes of some of the metastases. The incidence of node involvement in a series of 202 treated and 154 non-treated cases of cervical carcinoma, studied at autopsy, is set forth in Table I. Whereas, 58.5 per cent of the primary and 70 per cent of the secondary nodes were involved in the treated series, only 44.7 per cent of the primary and 39 per cent of the secondary nodes were affected in the non-treated series. Unfortunately, the correct interpretation of these differences is difficult, and it is possible only to theorize that the local extension was temporarily controlled in the treated cases, thus permitting time for the increase in nodal dissemination. That this theory is not completely valid is attested by the presence of distant metastases in 37.8 per cent of the treated and 32.5 per cent of the non-treated cases, an inappreciable difference in this relatively small series of cases. The true clinical stage of the disease is difficult to establish except at surgery or autopsy. Palpably normal parametria are not irrefutable evidence of the absence of malignant extension. Unfortunately, distant metastases do occur in the early stages of the disease. It is also possible, though relatively uncommon, to have distant metastases in the proved absence of parametrial involvement. However, no specific test or examination permits the demonstration of these distant metastases until their presence is subjectively manifested. The spread of cervical carcinoma by direct extension (Fig. 1) follows definite anatomical planes. The proximity of the bladder results in the frequent involvement of that organ. The cul-de-sac of Douglas serves as a barrier, delaying direct extension to the rectum until the disease has, in most instances, included the posterior vaginal wall. Lateral extension meets with little obstruction; thus parametrial involvement is common.