Objectives: This study was aimed to establish the most effective premature ovarian failure (POF) mouse model using Cyclophosphamide (CTX), busulfan (Bu), and cisplatin considering treatment duration of anticancer drugs and natural recovery time. Methods: POF was induced by intraperitoneally injecting CTX (120 mg/kg)/Bu (12 mg/kg) for 1 to 4 weeks or cisplatin (2 mg/ kg) for 3 to 14 days to C57BL/6 female mice aged 6 to 8 weeks. Controls were injected with equal volume of saline for the same periods. Body weight was measured every week, and ovarian and uterine weights were measured after the last injection of anticancer drug. To assess ovarian function, POF-induced mice were superovulated with pregnant mare serum gonadotropin and human chorionic gonadotropin, and then mated with male. After 18 hours, zygotes were retrieved and cultured for 4 days. Finally, the mice were left untreated for a period of times after the final injection of anticancer drug, and the time for natural recovery of ovarian function was evaluated. Results: After 2 weeks of CTX/Bu injection, ovarian and uterine weights, and ovarian function were decreased sharply. Cisplatin treatment for 10 days resulted in a significant decrease in ovarian and uterine weight, and ovarian function. When POF was induced for at least 2 weeks for CTX/Bu and for at least 10 days for cisplatin, ovarian function did not recover naturally for 2 weeks and 1 week, respectively. Conclusions: These results suggest that CTX/Bu should be treated for at least 2 weeks and cisplatin for at least 10 days to establish the most effective primary ovarian insufficiency mouse model. (J Menopausal Med 2018;24:196-203)