AIM: To investigate the effects of ovariectomy and estrogen replacement on learning and memory in mice in T-maze test. METHODS: Sixty female C 57 BL/6 J mice in three months old were randomly divided into sham and ovariectomized (OVX) groups. Each group had estrogen non-treated (Enon). Seven days treated (E 7 d) and forty days treated (E 40 d) subgroups. Each subgroup had ten mice. Mice were ovariectomized and raised until seven months old. 17 β-estradiol time release pellet (0.18 mg/pellet) was given i.c. seven days and forty days before T-maze were tested and then tested the numbers of correct choice in T-maze by mice. RESULTS: The correct choice in Enon subgroup, OVX group, significantly decreased d 1 was (42±10)% vs (54±9)%, d 2 was (47±11)% vs (62±10)%,and d 3 was (50±10)% vs (68±10)%. (P< 0.01, comparied to Enon subgroup in sham group). The correct choice in E 7 d and E 40 d subgroup, OVX group, significantly increased d 1 was (57±7)% vs (42±10)%, d 2 was (65±11)% vs (47±11)% and d 3 was (77±11)% vs (50±10)%. (P<0.01, comparied to Enon subgroup in OVX group). The effects of long-term administration of estrogen (E 40 d) in improving correct choice were better than short-term (E 7 d). There were no differences of correct choice between estrogen treated and non-treated subgroups in sham group (P> 0.05). CONCLUSION: Estrogen removal obviously damages learning and memory in mice and estrogen replacement can improve learning and memory in OVX mice. But over load of estrogen had no effects on processing of learning and memory in normal mice.