CA125, human epididymis secretary protein 4 (HE4) and the Risk of Ovarian Malignancy Algorithm (ROMA) could be used for diagnosing ovarian cancer (OCa). However, it has not been conclusively determined which of these markers yields the best diagnostic accuracy. Therefore, we conducted a meta-analysis to evaluate the diagnostic value of these markers. We systematically searched the PubMed and ScienceDirect databases and identified 32 studies that evaluated the role of CA125, HE4 and ROMA in diagnosing OCa. The bivariate random-effects approach was used to calculate the pooled estimates by considering the heterogeneity of major related parameters such as the menopausal status, International Federation of Gynecology and Obstetrics stages, detection method and blinded design. Three tests yielded similar discriminatory performances in the OCa diagnosis (AUC [95 % CI]-0.89 [0.86-0.92] for HE4; 0.87 [0.84-0.90] for CA125; 0.91 [0.88-0.93] for ROMA). HE4 yielded a higher specificity than CA125 and ROMA (HE4 93.60 [90.00-95.90] >CA125 82.10 [76.60-86.50] and ROMA 82.40 [77.40-86.50]), especially in the premenopausal subgroup (HE4 93.80 [88.40-96.80] >CA125 76.30 [63.30-85.70] and ROMA 85.10 [80.40-88.80]). In contrast, CA125 and ROMA performed significantly better in the postmenopausal subgroup than in the premenopausal subgroup (AUC [95 % CI]-CA125-premenopausal 0.85 [0.82-0.88]