STUDY QUESTION Can gonadotrophin-releasing hormone agonists (GnRH-a) preserve long-term fertility when administered prior to and concomitantly with chemotherapy? SUMMARY ANSWER GnRH-a display a differential protective effect on fertility, depending upon the specific chemotherapy-induced mechanism of ovarian injury. WHAT IS KNOWN ALREADY The role of GnRH-a in fertility preservation has been constantly debated and their use is considered experimental due to conflicting clinical evidence and paucity of data regarding their mechanism for ovarian protection. STUDY DESIGN, SIZE, DURATION In vivo model: 7-8 weeks old imprinting control region (ICR) mice were injected with GnRH-a (Leuprolide-acetate) or saline prior to and concomitantly with cyclophosphamide, doxorubicin or saline and sacrificed at various time-points on a longitudinal follow-up; 24 h (n = 36), 1 week (n = 40), 1 month (n = 36) and 9 months (n = 66) post chemotherapy treatment. Blood samples were drawn on Day 0 and on a monthly basis after chemotherapy treatment. On the day of sacrifice, blood samples were drawn and ovaries excised and processed for either immunohistochemistry (IHC), protein or RNA extraction. In vitro model: 21-23 days old Wistar-derived rats were sacrificed, their ovaries excised and primary granulosa cells (PGC) were either isolated for in vitro culture, or processed for immunofluorescence (IF) as well as for protein or RNA extraction. MATERIALS, SETTING, METHODS Ovarian reserve was estimated by serial measurements of serum anti-mullerian hormone (AMH), quantified by the AMH Gen II ELISA assay. Ovarian AMH and phosphorylated Akt (pAkt) were detected by immunoblotting. Vascular endothelial growth factor (VEGF) was measured by quantitative PCR. Ovarian GnRH receptor (GnRHR), AMH and CD34 were visualized by IHC, and apoptosis was evaluated using TdT (terminal deoxynucleotidyl transferase)-mediated dUDP nick-end labeling (TUNEL). MAIN RESULTS AND THE ROLE OF CHANCE Cyclophosphamide-induced ovarian injury caused a prompt decrease in AMH level (P < 0.01) and a further long-term decline in serum AMH (P = 0.017), indicating damage to the ovarian reserve. Pretreatment with GnRH-a diminished AMH-decrease (P < 0.05) and maintained serum AMH level in the long run (P < 0.05). Doxorubicin-exerted ovarian-vascular-injury is also displayed by an acute increase in ovarian VEGF level (P < 0.05) and a sustained decrease in serum AMH level (P < 0.001). This was followed by ovarian recovery manifested by increased neovascularization. GnRH-a delayed the recovery in AMH level and decreased the level of VEGF (P < 0.001), thus interfering with the vascular recovery subsequent to doxorubicin-induced vascular damage. LIMITATIONS, REASONS FOR CAUTION To portray the differential mechanism of each chemotherapy, cyclophosphamide and doxorubicin were given separately, whereas most of the clinical protocols include several types of chemotherapies. Thus, future study should explore a prospective evaluation of various chemotherapies, as well as combined chemotherapeutic protocols. WIDER IMPLICATIONS OF THE FINDINGS Our study demonstrates that different chemotherapy agents affect the ovary via diverse mechanisms and thus the administration of GnRH-a concomitantly, could be beneficial to a subpopulation of patients treated with cyclophosphamide-based protocols.