Background The predicting value of AMH for ovarian dysfunction after chemotherapy is controversial. This study is designed to evaluate the value of serum AMH clinically and theoretically. Patients, animals and methods We detected the serum estradiol, FSH and AMH in 144 pre-menopausal women with breast cancer receiving cyclophosphamide-based chemotherapy. The differences of hormones before and after chemotherapy were compared; the correlations among the hormones and amenorrhea, menstrual recovery were analyzed. In addition, serum AMH was detected randomly in 177 normal healthy women and 36 normal female C57BL/6J mice of different ages, meanwhile the status of ovary follicles was observed. Furthermore, 72 Balb/c nude mice with breast cancer were randomly assigned to three groups with different dosage of CTX (control, 100mg/kg, 200mg/kg), the alterations of serum AMH and ovary follicles were recorded and analyzed. Results Chemotherapy induced amenorrhea was associated with pre-chemo AMH level, E2 level and FSH level (P <0.0001). Recovery of menstruation was associated with pre-chemo AMH level (P <0.0001) but not E2 and FSH level (P >0.05). In breast cancer patients received chemotherapy, the serum AMH did not differ significantly between pre- and post- chemotherapy in patients younger than 35 years old (P>0.05), while dramatic decline was detected in patients over 35 years old (P<0.0001). In healthy women, the AMH level sharply declined after 35 years old ( P <0.0001) and remained relatively stable in early age. Similar results were obtained in normal mice experiments. The cancer-bearing mice exposed to 200mg/kg CTX endured significant decline of AMH levels and remarkable decrease of primordial and growing follicles (P <0.0001) Conclusion Our results indicate that AMH is effective for predicting post-chemo ovarian function only in premenopausal female breast cancer patients over 35 years old.