Candida esophagitis was diagnosed radiographically in 106 patients. Endoscopy was performed in 32 of these patients, and the diagnosis was confirmed in 27. The diagnosis was missed radiographically in seven other cases proved endoscopically. Thus, Candida esophagitis was diagnosed on esophagography in 80% (27/34) of proved cases, with five false-positive examinations. Single-contrast technique was employed in nine cases with a sensitivity of 55%, and double-contrast technique was used in 25 cases with a sensitivity of 88%. With double-contrast technique, esophagography appears to be a more accurate method for diagnosing Candida esophagitis than has been suggested in the literature.