To establish clinical and histologic determinants of survival, records of all UCLA patients with resectable melanoma metastatic to the lymph nodes during the years 1954-1976 were reviewed. These 150 patients were treated first with wide excision, lymphadenectomy, and with radiation/chemotherapy and/or additional surgery only if further recurrences developed. None received adjuvant immunotherapy or chemotherapy. In 97 of 139 patients with identified primary tumors, slides of the primary lesion were reviewed. Putative prognostic factors included age, sex, parity, site of primary tumor, presence of satellitosis, clinical status of nodes, histologic characteristics of primary lesion (Clark's level, thickness of tumor, presence/width of ulceration, and number of mitoses/HPF), time from biopsy of primary tumor to lymphadenectomy, and number of positive nodes. kaplan-Meier estimates of survival for the entire group at one, two, five, and ten years were 73, 55, 37, and 33%, respectively. Median follow-up period of survivors was four years. Univariate analyses using the log-rank test showed that thickness of the primary lesion (p less than 0.001), width of ulceration (p = 0.003), absence of ulceration (p = 0.024), and number of positive nodes (p = 0,.033) were prognostic for survival. In multivariate analysis by the Cox procedure, thickness of the primary (p = 0.001) and number of melanoma-containing nodes (p = 0.043) were prognostic for survival. Location of the primary tumor became marginally significant (p = 0.12) in the multrivariate model. These findings demonstrate the prognostic importance of characteristics of both the primary lesion and extent of regional dissemination. Future prospective randomized trials for (adjuvant) therapy of Stage II melanoma should be stratified by these variables.