OBJECTIVES: To compare several risk factors in the testicular biopsy of patients with pure seminoma with and without clinical metastasis at diagnosis. METHODS: We performed a retrospective study of patients with pure seminoma. The retroperitoneum was staged with computed tomography and the thorax with simple radiography and/or computed tomography, taking into account the original reports and clinical stage. The previous reports and original pathology plates were reviewed by pathologists who were unaware of the clinical stage of the patients. Patients with beta-human chorionic gonadotropin greater than 800 mUI/mL were excluded. RESULTS: A total of 86 patients had sufficient data and comprised the study cohort. Of the 86 patients, 62 had clinical Stage I (72%), 20 had Stage II (23%), and 4 had Stage III (5%). On univariate analysis, tumor size greater than 4 cm (P = 0.0135), testicular vascular invasion (P = 0.0042), rete testis invasion (P = 0.0002), tunica albuginea penetration (P = 0.00001), base of the spermatic cord invasion (P = 0.0002), epididymis invasion (P = 0.001), and vascular invasion of the cord (P = 0.024) were predictive of metastasis. On multivariate analysis, tumor size greater than 6 cm (odds ratio 6.9, 95% confidence interval 1.3 to 35, P = 0.02) and rete testis invasion (odds ratio 6.1, confidence interval 1.2 to 30, P = 0.025) remained as important predictors of metastasis (tumor size less than 6 cm was not significant on multivariate analysis). CONCLUSIONS: The results of this study have demonstrated that rete testis invasion and tumor size correlate independently with the presence of clinical metastasis at diagnosis of testicular seminoma.