OBJECTIVE To develop an experimental model of endometriosis in rats. METHODS Thirty adult female Wistar rats were used. The surgical technique consisted of median laparotomy with identification of the bicornuate uterus and resection of a 2-cm segment of the right uterine horn. A 0.25 cm(2) flap was removed from that structure and sutured to the abdominal wall with the endometrial side facing the peritoneal cavity. The rats were randomly divided into two groups according to the reoperation date: group 1 (n=15) was reoperated in 30 days, and group 2 (n=15), in 60 days. On the occasion of the second laparotomy, the implants were evaluated macroscopically, resected and referred for microscopic analysis with hematoxylin-eosin and immunohistochemical staining (HEMA, AE1 and AE2). RESULTS The implants developed in 83.3 % of group 1 and 71.4% of group 2. There was no statistically significant difference between the weights of the animals in the two groups. No statistically significant difference was found in the surface area of the induced lesions: in group 1, the mean was 0.37 cm(2) and in group 2, 0.25 cm(2). According to Keenan's semiquantitative histological classification (based on the preservation status of the epithelial layer of the endometrium), the mean for group 1 was 1.9 and for group 2, 2.4. CONCLUSIONS The technique used for inducing the development of endometriosis in rats was satisfactory.