Purpose: The American Joint Committee on Cancer (AJCC) recently proposed major revisions of the tumor-node-metastases (TNM) categories and stage groupings for cutaneous melanoma. Thirteen cancer centers and cancer cooperative groups contributed staging and survival data from a total of 30,450 melanoma patients from their databases in order to validate this staging proposal. Patients and Methods: There were 17,600 melanoma patients with complete clinical, pathologic, and follow-up information. Factors predicting melanoma-specific survival rates were analyzed using the Cox proportional hazards regression model. Follow-up survival data for 5 years or longer were available for 73% of the patients. Results: This analysis demonstrated that (1) in the T category, tumor thickness and ulceration were the most powerful predictors of survival, and the level of invasion had a significant impact only within the subgroup of thin (≤ 1 mm) melanomas; (2) in the N category, the following three independent factors were identified: the number of metastatic nodes, whether nodal metastases were clinically occult or clinically apparent, and the presence or absence of primary tumor ulceration; and (3) in the M category, nonvisceral metastases was associated with a better survival compared with visceral metastases. A marked diversity in the natural history of pathologic stage III melanoma was demonstrated by five-fold differences in 5-year survival rates for defined subgroups. This analysis also demonstrated that large and complex data sets could be used effectively to examine prognosis and survival outcome in melanoma patients. Conclusion: The results of this evidence-based methodology were incorporated into the AJCC melanoma staging as described in the companion publication. © 2001 by American Society of Clinical Oncology.