To investigate cyclooxygenase (COX- 2) expression within different endometriotic lesions and to assess whether these expression patterns correlate with clinical characteristics. Design: Retrospective cross-sectional study. Setting: University Hospital. Patients: Seventy patients with histologically confirmed exclusively peritoneal (n = 20), ovarian (n = 19) or deep-infiltrating (n = 31) endometriosis and a detailed medical history. Intervention: Immunohistochemical analysis for COX-2 was performed on 108 endometriotic lesions. Measurements and main results: COX- 2 intensity, percentage of stained glandular endometriotic cells, and correlation of COX- 2 expression with clinicopathological parameters. Semiquantitative COX- 2 expression did not differ between distinct morphological types of endometriosis and showed no association with the menstrual cycle. Patients with peritoneal-only endometriosis suffering from moderate or severe chronic pelvic pain showed significantly more frequent COX- 2 over expression than asymptomatic patients or patients with minimal symptoms. In patients with exclusively ovarian or deep-infiltrating endometriosis no association between COX- 2 expression and clinical parameters, such as chronic pelvic pain, dysmenorrhoea, dyspareunia, sterility, lower urinary tract symptoms or gastrointestinal symptoms was observed. Conclusion: Peritoneal endometriotic lesions with increased COX- 2 expression have a special relevance for the development of chronic, nonmenstruation-associated, pelvic pain in endometriotic patients. These patients may benefit from therapy with COX- 2 inhibitors. To investigate cyclooxygenase (COX-2) expression within different endometriotic lesions and to assess whether these expression patterns correlate with clinical characteristics. Design: Retrospective cross-sectional study. Settings: University Hospital. Patients: Seventy patients with histologically confirmed exclusively peritoneal (n = 20), ovarian (n = 19) or deep-infiltrating (n = 31) endometriosis and a detailed medical history. Intervention: Immunohistochemical analysis for COX- 2 was performed on 108 endometriotic lesions. Measurements and main results: COX- 2 intensity, percentage of stained glandular endometriotic cells, and correlation of COX-2 expression with clinicopathological parameters. Semiquantitative COX- 2 expression did not differ between distinct morphological types of endometriosis and showed no association with the menstrual cycle. Patients with peritoneal-only endometriosis suffering from moderate or severe chronic pelvic pain showed significant more frequent CO X-2 over expression than asymptomatic patients or patients with minimal symptoms. In patients with exclusively ovarian or deep-infiltrating endometriosis no association between COX-2 expression and clinical parameters, such as chronic pelvic pain, dysmenorrhoea, dyspareunia, sterility, lower urinary tract symptoms or gastrointestinal symptoms were observed. Conclusion: Peritoneal endometriotic lesions with increased COX-2 expression have a special relevance for the development of chronic, nonmenstruation-associated, pelvic pain in endometriotic patients. These patients may benefit from therapy with COX-2 inhibitors.