Objective To study the therapeutic effect of recombinant adeno-associated virus carrying human endostatin gene therapy on endometriosis in mice model. Methods Recombinant adeno-associated virus vector carrying human endostatin gene and enhanced green fluorescent proteins gene (rAAV2-endostatin-EGFP) was constructed. Endometrium was from 12 patients with leiomyoma undergoing hysterectomy in Second Hospital, Tianjin Medical University between November and December 2008. Endometriosis models of nude mice were established by transplanting human endometrial fragments intooperitoneal surface. After 1 week, those 60 mice were divided into 3 groups: treatment group including 20 mice injected with rAAV2-endostatin-EGFP to ectopic lesion, control group including 20 mice injected with rAAV2-EGFP to ectopic lesion and blank control group including 20 mice injected with phosphate buffered saline (PBS) to the ectopic lesion. At 1, 2 and 3 weeks after treatment, those mice underwent laparotemy to observe the location and size of ectopic lesion in abdominal cavity. The expression of endostain protein, number of gland, microvessel density (MVD) and vascular endothelial growth factor (VEGF) were measured in ectopic lesions. The serum level of estradiol and progesterone were detected in nude mice among every groups. Results (1) All endometriosis of nude mice models were established successfully through peritoneum transplanting. After 1 week's treatment, flat lesion nodes, decreased gland number and narrow and atrophy glandular cavity were observed by light microscope. (2) The endostatin gene was transferred into nude mice successfully and expressed effectively. It was observed that endostatin protein expression was shown with enhanced green fluorescent proteins in ectopic lesion. (3) Glands number of ectopic lesion in rAAV2-endostatin-EGFP group(7.8±1.9,7.0±1.5 and 5.5±1.7) were significantly less than 10.1± 1.7, 10.2±2.0 and 9.8±2.4 in rAAV2-EGFP control group and 10.2±2.2,10.0±2.0 and 9.7±2.2 in PBS control group at 1,2 and 3 weeks after treatment(all P<0.05). Glands number of ectopic lesion in rAAV2-endostatin-EGFP group at 3 weeks was significantly less than those at 1 and 2 weeks after treatment (P<0.05). (4) MVD of ectopic lesion in rAAV2-endostatin-EGFP group (12.2±1.5,11.4±2.1 and 9.0±1.4) was significantly less than those at rAAV2-EGFP control group (16.5±1.7,16.5±1.9 and 16.9±1.9) and PBS control group (16.2±1.6,16.0±1.6 and 16.3±1.7) at 1,2 and 3 weeks after treatment (all P<0.05) . MVD of ectopic lesion in rAAV2-endostatin-EGFP group at 3 weeks was significantly less than those at 1 and 2 weeks after treatment(P<0.05). (5) The rate and density of VEGF expression at ectopic lesion in rAAV2-endostatin-EGFP group (35%, 30%, 25% and 1.60±0. 43,1.33± 0. 30,1.03±0.36) were significantly less than those at rAAV2-EGFP control group (80% ,75% ,85% and 2.43±0.53,2.43±0.29,2.66±0.45) and PBS control group (85% ,90% ,90% and 2.36±0.53,2.64± 0.57,2.53±0.52) at one 1, 2 and 3 ,weeks after treatment (all P<0.05). The expression of VEGF at ectopic lesion in rAAV2-endostatin-EGFP group at 3 weeks was significantly less than those at 1 and 2 weeks after treatment (P<0.05). (6) The level of estradial and progesterone in serum of nude mice of rAAV2-endostatin-EGFP group [ E_(2)> : (48±7 ) pmol/L, P: (61±8 ) nmol/L ] did not reach statistical difference when compared with those at rAAV2-EGFP control group [ E_(2): (50±9) pmol/L, P: (60±10) nmol/L] and PBS control group [E_(2):(48±7)pmol/L,P: (58±10)nmol/L,P>0.05]. Conclusions The recombinant adeno-asseciated virus carrying human endostatin gene therapy could inhibit angiogenesis at endometriotic lesions and not influence steroid level. The antiangiogenic gene therapy might become a novel option for endometriosis.