Obective: To evaluate the efficacy of five methods: pelvic examination (PE), transvaginal ultrasonography (US), Doppler ultrasonography, serum CA 125 assay and serum CA 72-4 assay, alone or associated, to predict malignancy in patients presenting a pelvic mass originating in the ovary. Methods: 92 patients underwent a standard protocol for physical examination, CA 125, CA 72-4, transvaginal ultrasonography and Doppler ultrasonography. Results: Eighteen women were dropped from the study because they had clearly benign masses; two women were dropped from the study because they had clearly malignant lesions. Twenty-two malignant (30%) and 50 benign (70%) pelvic tumors were found When one method was considered alone the best sensitivity (SENS) was found in physical examination (90%) and the best specificity (SPEC) was found in CA 72-4: 88%. If all indicators were positive, the SPEC was 100% but the SENS was 40%. Logistic regression analysis prediction of the character of the pelvic masses was correct in 86%. Conclusion: Some additional information to discriminate between malignant and benign pelvic masses can be obtained from the valuation of serum tumor markers, particularly CA 72-4. Also Doppler ultrasonography appeared to be useful in the differential diagnosis of pelvic tumors. The prediction of the character of the pelvic masses calculated by a logistic model in which PE, US, CA 125, and CA 72-4 are included is very good.