Background <p>Criteria for malignancy in smooth muscle tumors vary by anatomic site. However, establishment of the primary site is not always possible preoperatively in the pelvis and retroperitoneum of female patients. Estrogen receptor (ER) expression rates in non-gynecologic smooth muscle tumors remain a subject of debate. We sought to study a cohort of well-characterized neoplastic and non-neoplastic non-gynecologic smooth muscle in both sexes to explore the utility of ER status in determination of site of origin.</p> Design <p>A retrospective search was performed for cases of non-gynecologic leiomyosarcoma. Only cases where the site of origin could be confirmed by radiologic features, intraoperative findings, and/or gross description were included. Non-neoplastic non-gynecologic smooth muscle tissues were also collected. A representative section from each specimen was immunostained for ER. ER expression was scored as follows: &lt; 1% (0), 1–25% (1 +), 26–50% (2 +), &gt; 50% (3 +). Specimens were considered ER positive if ≥ 1% of tumor cells showed at least moderate nuclear staining for the antibody. Clinical data, including follow-up, was collected.</p> Results <p>Sixty-six cases of non-gynecologic leiomyosarcoma (36 females, 30 males) were identified, arising from large/medium blood vessels of the retroperitoneum/pelvis (23), subfascial soft tissue of the extremities, chest or trunk (17), suprafascial soft tissue of the extremities, chest or trunk (11), genitourinary tract (6), and gastrointestinal tract (9). FNCLCC grading was as follows: Grade 1 in 10 cases (15.2%), Grade 2 in 39 cases (59.1%), and Grade 3 in 17 cases (25.8%). ER expression was detected in 11 of 66 (17%) of non-gynecologic leiomyosarcoma. Follow-up was available in 64 patients (0 to 163&#xa0;months, median 23&#xa0;months); 6 patients developed local recurrence, 26 patients developed distant metastasis, and 12 patients died of disease. When cases were categorized into two groups based on gender, 9/36 (25.0%) of female tumors were ER-positive compared to 2/30 (6.7%) of male cases, and the difference was statistically significant (Fisher's exact test, <i>p</i> = <i>0.046</i>). Leiomyosarcomas arising from large/medium vessels of the retroperitoneum/pelvis were significantly more likely to express ER (7/23, 30.4%) than those from other non-gynecologic sites (4/43, 9.3%) (Fisher's exact test, <i>p</i> = <i>0.035</i>). There was no correlation between ER status and FNCLCC grade or outcome. Out of a total of 43 non-neoplastic smooth muscle tissues, 14 (33.3%) tissue samples were positive for ER expression. Analysis by site showed that non-neoplastic smooth muscle arising from retroperitoneal/pelvic vessels was significantly more likely to express ER (10/16, 62.5%) compared to other non-gynecologic sites (4/27, 14.9%) (Fisher's exact test, <i>p</i> = <i>0.002</i>).</p> Conclusion <p>ER expression may be present in both neoplastic and non-neoplastic smooth muscle at non-gynecologic sites. Extra-gynecologic leiomyosarcoma arising in female patients and from retroperitoneum/pelvic vasculature are more likely to show ER expression compared to male patients and other non-gynecologic sites, respectively. These findings limit the utility of ER expression in determining the site of origin when evaluating smooth muscle neoplasms of the pelvis and retroperitoneum.</p>