Endoscopic ultrasound (EUS) allows visualization of celiac lymph nodes (CLNs) and celiac ganglia (CG). Reliably distinguishing these structures is important for tumor staging and CG ablative therapies. We aimed to evaluate the accuracy of EUS in distinguishing CLNs from CG using a strict cytopathology reference standard. We also determined the rate of detection of CLN and CG by conventional cross-sectional imaging. From EUS and cytopathology databases, we identified all patients who underwent EUS-FNA of a presumed CLN or CG from October 1, 2004, through March 1, 2017, and compared the findings with those from cytology (reference standard). Indeterminate cytology results were re-reviewed. EUS imaging (ie, index test) results were compared with those from the reference standard. An expert radiologist re-reviewed computed tomography and magnetic resonance images from 100 lesions, from 94 randomly selected patients with a reference standard, to determine the rates of CLN and CG detection. A total of 504 patients (mean age, 63.4 ± 13.2 years; 292 men) underwent a median of 7 EUS-FNA passes (range, 1-13) for a total of 566 lesions perceived to be either a CLN or CG; the cytology reference standard was available for 521 lesions (92.1%). When we excluded indeterminate cytology results, the EUS accurately identified 281/286 CLNs (98.3%) and 166/186 CGs (89.2%), for an overall accuracy of 447/472 (94.7%). EUS-FNA distinguished CG from CLNs with a 93.3% sensitivity, 93.7% specificity, a positive predictive value of 96.2%, and a negative predictive value of 89.2%. Of 100 lesions in 94 patients randomly selected for a second expert radiology review, computed tomography and magnetic resonance imaging detected 59/67 CLNs (88.1%) and 13/33 CG (39.4%). EUS accurately distinguishes CLNs from CG. EUS might therefore be used to increase the accuracy of tumor staging, to select tumor stage-appropriate therapy, and to guide CG-ablative therapies.