Background: Endoscopic ultrasound (EUS)-guided sampling has been widely used for pathologic diagnosis of pancreatic lesions and intra-abdominal lymphadenopathy. However, its effectiveness for diagnostic decision making in indeterminate radiological diagnosis has not been well determined. Materials and Methods: From March 2012 to October 2015, 98 consecutive patients who underwent EUS-guided FNA for solid intra-abdominal lesions were retrospectively analyzed (100 procedures). The purpose of EUS-guided sampling was classified as (1) confirmation of a high-confidence radiological diagnosis (High-confidence group) or (2) decision making in the differential diagnostic dilemma for indeterminate radiological diagnosis (Indeterminate group). The accuracies of EUS-guided sampling according to the purpose were analyzed and then compared. Results: Of the 100 procedures, 22 procedures (22%) came under the Indeterminate group, whereas 78 came under the High-confidence group. The accuracies did not differ between the Indeterminate and the High-confidence groups (86.4% vs. 88.5%, p = 1.000). Clinical conditions that required EUS-guided sampling for indeterminate radiological diagnosis were (1) pancreatic cancer vs. benign disease (n = 8; e.g., pancreatic cancer vs. mass-forming pancreatitis), (2) recurrence of previous/pre-existing cancer vs. benign disease (n = 5; e.g., recurrent gastric cancer vs. reactive lymph node), (3) pathologic differentiation of presumed malignancy (n = 6; e.g., lymphadenopathies in the previous history of esophageal cancer and colon cancer), or (4) miscellaneous (n = 3; e.g., tuberculous lymphadenopathy vs. other condition). Conclusions: EUS-guided sampling demonstrated an accuracy of 86.4% in the clinical setting of indeterminate radiological diagnosis, which was not different from that of the confirmation of high-confidence diagnosis.