Abstract Study question How does freezing, in vitro culture (IVC) and grafting to chorioallantoic membrane (CAM) affect follicle activation through the PI3K pathway in human ovarian tissue? Summary answer Freezing does not trigger further follicle activation. Although IVC and CAM do induce follicle activation, the process is significantly curbed in the CAM group. What is known already While massive follicle activation has been shown to occur through the PI3K pathway in frozen-thawed human ovarian tissue after both short-term xenotransplantation and IVC, it remains poorly understood in fresh ovarian tissue, raising many questions about the effect of freezing on follicle activation. Furthermore, grafting ovarian tissue to CAM has been found to limit follicle activation in animal ovarian tissue, though its impact on human follicles is yet to be elucidated. Study design, size, duration Thirty fresh cortical fragments from 5 patients and 50 frozen-thawed tissue pieces from another 5 patients were investigated. We first compared fresh fragments (fresh-IVC) with frozen-thawed tissue (FT-IVC) by IVC. These cultured frozen-thawed fragments were then examined against frozen-thawed fragments grafted to CAM (FT-CAM). After both IVC and CAM grafting, ovarian cortical pieces (4x2x1 mm) were analyzed on day (D) 0, 1 and 6. Participants/materials, setting, methods Follicle analyses included histology (count and classification), transmission electron microscopy (ultrastructure), immunohistochemistry (Ki67 for follicle proliferation, c-caspase-3 for follicle apoptosis, LC3B for follicle autophagy, p-AKT for PI3K activation and p-rpS6 for PI3K/mTOR activation), and immunofluorescence (FOXO1 for PI3K/FOXO activation). Subcellular localization of FOXO1 was determined in primordial follicles on high-resolution images using structured illumination microscopy. Droplet digital PCR further investigated expression of PI3K pathway-related genes (TSC1 and CDKN1B) and oocyte growth-related genes (GDF9 and LHX8). Main results and the role of chance No differences were detected between the fresh-IVC and the FT-IVC groups in any conducted analyses. After both IVC and CAM grafting, a significant decrease was observed in primordial follicle proportions in all groups on D6 vs D0 (p < 0.01), associated with a substantial increase in intermediate follicle proportions (p < 0.05). In the FT-CAM group, however, the percentage of primordial follicles stabilized between D1 and D6, showing significantly higher rates than the FT-IVC group on D6 (p = 0.04) and reflecting better preservation of the primordial follicle pool in the FT-CAM group. While apoptotic follicle rates increased in all conditions (D0 vs D6, p < 0.03), they remained significantly lower in the FT-CAM group than in the FT-IVC group on D6 (p = 0.01), confirming superior follicle preservation in CAM-grafted tissue. In primordial follicles, significantly lower levels of p-rpS6 were noted over time in all conditions (D0 vs D6, p < 0.009), but there was a significant shift in FOXO1 to a nuclear localization in the FT-CAM group compared to the FT-IVC group on D1 (p = 0.03), indicating inhibition of follicle activation. Limitations, reasons for caution As demonstrated by our results, the CAM model appears to be effective at preserving the follicle pool. However, our study did not investigate the underlying mechanisms behind this improved follicle preservation after thawing. Wider implications of the findings Our data confirm that the freezing procedure itself does not cause follicle burnout. Moreover, the CAM model looks to be capable of preserving the follicle pool, hypothetically thanks to rapid revascularization and circulating embryonic anti-Müllerian hormone. Trial registration number Not applicable