From 1977 to 1984 the Gynecologic Oncology Group (GOG) conducted a prospective clinical and surgical staging protocol of squamous cell carcinoma of the vulva (n = 637). The patients with superficial (5 mm or less invasion) lesions were the subject of a previous report (n = 272). The subject of this report is on factors that predict groin node metastasis based on all 588 evaluable patients. Comparisons between the two reports are made. Almost half of this group (49.3%) had minimal tumor thickness (< or = mm). Almost one-third of patients had small vulvar lesions (< or = cm). Groin node metastasis was 18.9% for the < or = 2-cm diameter tumors and 41.6% for the > 2-cm diameter lesions. The inaccuracy of clinical palpation of the groin nodes (23.9% false negative) largely accounts for underestimation of extent of disease. Body weight was not related to the sensitivity of detecting positive groin nodes (P = 0.26). Using the logistic model, independent predictors of positive groin nodes were identified (in order of importance): less tumor differentiation by GOG criteria (P < 0.0001), suspicious or fixed/-ulcerated nodes (P < 0.0001), presence of capillary-lymphatic involvement (P < 0.0001), older age (P = 0.0002), and greater tumor thickness (invasion) (P = 0.03). Lesion size and location were not independent predictors of positive groin nodes.