It is broadly accepted that the false positive (FP) rate for endoscopic ultrasound fine needle aspiration (EUS FNA) is 0-1%. It was hypothesised that the FP and false suspicious (FS) rates for EUS FNA are greater than reported. A study was undertaken to establish the rate and root cause of discordant interpretation. Using a prospectively maintained endoscopic database, cytohistological discordant EUS FNA examinations from 30 July 1996 to 31 December 2008 were identified retrospectively. Tertiary referral centre. Discordant FNA was defined by positive or suspicious FNA cytology in the absence of malignancy or neoplasm in the subsequent surgical pathology specimen, specifically in the absence of neoadjuvant therapy. Three cytopathologists conducted a blinded review of randomised discordant and matched specimens. FNA was performed in 5667/18 066 (31.4%) patients undergoing EUS, of whom 2547 had cytology results interpreted as 'positive' or 'suspicious' or 'atypical' for malignancy or neoplasm. Subsequent surgical resection without prior neoadjuvant therapy was performed in 377 patients with positive or suspicious cytology. The FP rate was 20/377 (5.3%) and increased to 27/377 (7.2%) when FS cases were included. The incidence of discordance was consistent over time (1996-2002: 10/118 (8.6%) vs 2003-2008: 17/259 (6.6%); p=0.5) and was higher in non-pancreatic FNA (15%) than pancreatic FNA (2.2%; p=0.0001). Two-thirds of the non-pancreatic FP cases involved sampling of perioesophageal or perirectal nodes in patients with luminal neoplasms or Barrett's oesophagus. Following pathological re-review, discordance was attributed to translocated cell contamination/sampling error (50%) or cytopathologist interpretive error (50%). These findings refute the accepted paradigm that FP cytology rarely occurs with EUS FNA. Further investigation revealed that FP FNA developed secondary to endosonographer technique or initial cytological misinterpretation, and is particularly likely when perioesophageal or perirectal nodes are aspirated in the setting of a luminal neoplasm or Barrett's oesophagus. Further study is needed to determine the significance of these findings and potential impact on the performance of FNA and patient outcomes.