To develop a predictive model for ovarian failure (OF) after chemotherapy in young post-pubertal women with cancer. Retrospective, monocentric cohort study including 348 patients referring to the Oncofertility Unit of San Raffaele Hospital (Milan, Italy) from August 2011 to January 2020. A predictive model was constructed by multivariate logistic regression and receiver operating characteristic analysis. Data about menstrual function resumption were available for 184 patients. The best predictive model for OF was identified by the combination of age; number of chemotherapy lines; vincristine, adriamycin, ifosphamide/adriamycin, ifosphamide; capecitabine; adriamycin, bleomycine, vinblastine, doxorubicin (area under the curve = 0.906; CI 95% 0.858–0.954; p = 0.0001). The model predicts the probability of loss of ovarian function at cancer diagnosis and with every change of treatment.