Background. Duodenal gastrointestinal stromal tumors (GISTs) are a small subset of GISTs, and their management is poorly defined. We evaluated surgical management and outcomes of patients with duodenal GISTs treated with pancreaticoduodenectomy (PD) versus local resection (LR) and defined factors associated with prognosis. Methods. Between January 1994 and January 2011, 96 patients with duodenal GISTs were identified from five major surgical centers. Perioperative and long-term outcomes were compared based on surgical approach (PD vs LR). Results. A total of 58 patients (60.4 %) underwent LR, while 38 (39.6 %) underwent PD. Patients presented with gross bleeding (n = 25; 26.0 %), pain (n = 23; 24.0 %), occult bleeding (n = 19; 19.8 %), or obstruction (n = 3; 3.1 %). GIST lesions were located in first (n = 8, 8.4 %), second (n = 47; 49 %), or third/fourth (n = 41; 42.7 %) portion of duodenum. Most patients (n = 86; 89.6 %) had negative surgical margins (R0) (PD, 92.1 vs LR, 87.9 %) (P = 0.34). Median length of stay was longer for PD (11 days) versus LR (7 days) (P = 0.001). PD also had more complications (PD, 57.9 vs LR, 29.3 %) (P = 0.005). The 1-, 2-, and 3-year actuarial recurrence-free survival was 94.2, 82.3, and 67.3 %, respectively. Factors associated with a worse recurrence-free survival included tumor size [hazard ratio (HR) = 1.09], mitotic count[10 mitosis> 50 HPF (HR = 6.89), AJCC stage III disease (HR = 4.85), and NIH high risk classification (HR = 4.31) (all P<0.05). The 1-, 3-, and 5-year actuarial survival was 98.3, 87.4, and 82.0%, respectively. PD versus LR was not associated with overall survival. Conclusions. Recurrence of duodenal GIST is dependent on tumor biology rather than surgical approach. PD was associated with longer hospital stays and higher risk of perioperative complications. When feasible, LR is appropriate for duodenal GIST and PD should be reserved for lesions not amenable to LR.