Background: The purpose of this study is to prospectively evaluate the performance characteristics of endoscopy and EUS in the diagnosis of GI subepithelial masses. Methods: A total of 100 consecutive patients referred for the evaluation of a suspected GI subepithelial lesion were prospectively studied with endoscopy followed by EUS. Size, color, mobility, location (intramural or extramural), consistency (solid, cystic, or vascular), and presumptive diagnosis were recorded at the time of endoscopy. EUS then was performed, and size, echogenicity, location, and presumptive diagnosis were determined. Results: A total of 100 subepithelial lesions were evaluated. Endoscopy had 98% sensitivity and 64% specificity in identifying intramural lesions. Size measurement by endoscopy correlated with size measurement by EUS (r = 0.88). Histology was obt-ained in 23 cases, with the presumptive EUS diagnosis correct in only 48% of cases. Most incorrect EUS diagnoses occurred with hypoechoic 3rd and 4th layer masses. Conclusions: Endoscopy has high sensitivity but low specificity in identifying the location (intramural or extramural) of subepithelial lesions. In addition, EUS imaging alone is insufficient to accurately diagnose 3rd and 4th layer hypoechoic masses, and histologic confirmation should be obtained whenever possible. Background: The purpose of this study is to prospectively evaluate the performance characteristics of endoscopy and EUS in the diagnosis of GI subepithelial masses. Methods: A total of 100 consecutive patients referred for the evaluation of a suspected GI subepithelial lesion were prospectively studied with endoscopy followed by EUS. Size, color, mobility, location (intramural or extramural), consistency (solid, cystic, or vascular), and presumptive diagnosis were recorded at the time of endoscopy. EUS then was performed, and size, echogenicity, location, and Endoscopy had 98% sensitivity and 64% specificity in identifying intramural lesions. Size measurement by endoscopy correlated with size measurement by EUS (r = 0.88). Histology was obt- ained in 23 cases, with the presumptive EUS diagnosis correct in only 48% of cases. Most incorrect EUS diagnoses occurred with hypoechoic 3rd and 4t h layer masses. Conclusions: Endoscopy has high sensitivity but low specificity in identifying the location (intramural or extramural) of subepithelial lesions. In addition, EUS imaging alone is insufficient to accurately diagnose 3rd and 4th layer hypoechoic masses, and histologic confirmation should be obtained whenever possible