Objectives: The aim of this study was to compare and evaluate the quality of CA125, HE4, logistic regression model based on CA125 and HE4, and ROMA algorithm in preoperational differential diagnostics of the ovarian tumors. Material and methods: To the study 110 patients enrolled. Based on histopatological examination of removed tumors, they were divided into study group (56 cancer patients) and control one (nonmalignant 54 patients). Serum CA125 and HE4 concentrations were measured following a standard procedure. Results: A commonly accepted referential value for CA125 is 35 IU/ml. In our study, this cut-off value yielded very low sensitivity and specificity results (85.2% and 63.6%, respectively). When we adopted HE4 normal value to be 140 pM,the sensitivity and specificity obtained in the investigated population was 68.5% and 94.6%, respectively. When the cut-off value for HE4 was adopted as 74 pM, the sensitivity improved considerably (88.9%), but specificity decreased to 85.7%. In case of CA125 when we adopted Ca125 normal value to be 77 IU/ml, the sensitivity and specificity obtained in the investigated population was 81.5% and 83.6%, respectively. In analysis based on combination of biomarkers, the highest sensitivity was obtained for the logistic regression model based on CA125 and HE4 (89.5%). A little bit lower sensitivity was achieved for HE4 used as a single diagnostic test (88.9%). The highest specificity was observed for ROMA algorithm (94.5%). This means that ROMA algorithm is the best diagnostic tool to differentiate between the malignant and non-malignant ovarian tumors. Conclusions: 1. ROMA algorithm yielded the highest specificity and slightly lower sensitivity in the case of differential diagnosis between malignant and non-malignant ovarian tumors. Therefore, it should become a basic tool in the ovarian tumors diagnosis prior to a surgery. 2. HE4 as a single diagnostic test (based on one marker) was found to be better suited to the ovarian tumor differential diagnosis than CA125 test. 3. Combined test, based on double marker analysis, should be applied and then the risk of the ovarian cancer should be calculated. This approach is more effective than single marker analysis.