36 Background: GIST is the most common mesenchymal tumor of the gastrointestinal tract although it constitutes only 1 percent of primary GI malignancy. Some GIST develops outside the alimentary canal such as omentum, mesentery, and retroperitoneum. These types are called extra-gastrointestinal stromal tumors (EGISTs). In this study, we retrospectively analyzed the clinicopathologic features of EGIST patients from 4 institutes in Korea Methods: From 2004 to 2012, a total of 43 patients with EGIST were identified. The clinicopathologic features including gender, age, location, tumor size, hisology, mitotic rate, IHC features, and genetic status were collected along with survival analysis. Results: The median age was 55 (range 29-80) and M:F ratio was 20:23. The most common sites were mesentery (11) and retroperitoneum (11) followed by omentum (7) and vagina (3). Other primary sites were liver (2), ovary (2), pancreas (2), perianal area (2), chest wall (1), pleura (1) and prostate (1). The median tumor size was 8.0 cm (range 2.6 – 30) and median mitotic rate was 5.0/50HPF (0 – 185) 4 cases were negative for c-Kit IHC, which were all confirmed by either IHC for DOG1 or mutational analysis. KIT analysis was performed in 10 cases with 6 wild-types and 4 exon 11 mutation were found. As for PDGFRα, 4 cases were analyzed which were all exon 18 mutation. Among 43 patients, 26 patients had undergone curative resection and 8 patients had unresectable disease and had taken palliative imatinib. Among 26 patients who had undergone curative resection, 12 patients had taken adjuvant imatinib and 14 patients had been observed without imatinib. The median tumor size of two groups were 11.0 (3.0 – 19.0) vs. 5.85 (2.6 – 18.9) and the median mitotic rate was 7.0 (2 – 18.5) vs. 1.0 (0 – 45). Regarding the 9 patients taken palliative imatinib, median tumor size was 20.9 (16.5 – 30.0), median mitotic rate was 15.0 (3 – 60) and median PFS was 22.7 months (95% CI 7.7 – 37.7). Conclusions: As the biologic behavior of GISTs is different by location, a more stratified strategy is needed for managing EGISTs including incorporation of molecular features. We are planning to perform further mutation analysis in this cohort.