Groin dissection was performed in 158 patients with malignant melanoma (superficial dissection, 76 patients; radical dissection, 82 patients). Of 63 patients with palpable nodes, 57 patients (90%) had histologic involvement. Of 93 patients with nonpalpable nodes, 31 patients (33%) had histologically positive nodes. The 5-year survival rate for patients with histologically negative nodes (n = 69) was 77%; the 5-year survival rate for patients with histologically positive nodes (n = 89) was 43%. The respective 5-year disease-free survival rates were 72% and 34%. Of 57 patients with palpable, positive inguinal nodes, 21 patients (37%) had involvement of the deep nodes. Of 31 patients with nonpalpable, histologic involvement of the inguinal nodes, six patients (19%) had or developed involvement of the deep nodes. One of two patients with uncertain clinical status of the nodes preoperatively had positive deep nodes. In prophylactic node dissection, frozen section of the inguinal group of the nodes does not provide a reliable method, because of sampling errors, in determining microscopic involvement of the nodes and in deciding whether a superficial or radical groin dissection is to be done. For patients with positive nodes the 5-year survival rate was 48% when only the inguinal group was involved and was 28% when both inguinal and deep nodes were involved; the respective 5-year disease-free survival rates were 39% and 20%. Survival after therapeutic groin dissection may partly depend on the thoroughness of the procedure. Patients who have positive, deep nodes and who are undergoing an incontinuity dissection of the inguinal, iliac, and obturator nodes have an appreciable 5-year survival rate.