This is a retrospective study of patients treated for early-stage cervical cancer to identify pathologic risk factors associated with ovarian metastases and, therefore, to establish when ovarian preservation can be performed without increasing the risk of relapse in order to improve the quality of life in premenopausal patients. Between 1982 and 2004, 1965 patients with FIGO stage IA2-IB-IIA cervical squamous cell carcinoma and nonsquamous histology types were surgically treated; 1695 (86%) patients underwent primary radical hysterectomy, bilateral salpingo-oophorectomy, and pelvic node dissection, the remaining 270 patients (14%) had their ovaries preserved. The clinical records were reviewed for all patients and clinical features at presentation, the histopathology and follow-up data were recorded. Overall, ovarian metastases were diagnosed in 16 of 1695 patients, for an incidence rate of 0.9%. Univariate analysis shows age (</=45 vs >45 years: P = 0.0079), FIGO stage (IB1-IIA </=4 cm vs IB2-IIA >4 cm: P = 0.0133), histology (squamous vs nonsquamous, P = 0.0014), noninvolved peripheral stromal thickness (<3 vs >3 mm: P = 0.0001), lymphvascular space involvement (present vs absent, P = 0.0007), lymph node status (positive vs negative, P = 0.00009) to be statistically associated with the presence of ovarian metastases. Multivariate analysis shows only age (P = 0.0119), FIGO stage (P = 0.011), histology (P = 0.001), and unaffected peripheral stromal thickness (<3 mm: P = 0.037) to be independent risk factors for ovarian metastases. Based on the present data and on the data available in the literature, ovarian preservation could be safely performed in young patients with early-stage squamous cell carcinoma (histology as the most significant risk factor), with macroscopically normal ovaries, and with preserved peripheral unaffected cervical stroma.