Background: Endoscopic ultrasound (EUS) has been regarded as the most accurate modality for locoregional staging of pancreatic malignancy. However, several recent studies have questioned this. The current study assessed the accuracy of EUS in determining preoperative resectability of pancreatic neoplasia. Methods: A retrospective review was performed of patients with pancreatic malignancy who had preoperative EUS and underwent surgery. EUS-predicted resectability was compared with surgical resectability. Where available, accuracies of vascular and nodal staging were also assessed. Results: Forty-five patients were identified (mean age = 60 years, age range = 36–79 years). All patients underwent surgical exploration; vascular staging was available in 32 cases and 17 cases underwent surgical resection. The sensitivity, specificity, and accuracy of EUS in determining unresectability were 66%, 100%, and 78% respectively. Overall EUS stage concurred with surgical stage in 56%, greater than surgical stage in 4%, and less than surgical stage in 40%. Vascular staging on EUS had a sensitivity of 69% and a specificity of 100%. Accuracy of nodal staging was 71%. Conclusion: EUS had a high specificity for assessing unresectable pancreatic malignancy. This technique should be used to avoid unnecessary surgical exploration of incurable lesions. However, EUS had only a moderate sensitivity, and a proportion of patients staged preoperatively as having resectable disease will not be surgically resectable.