Background: If areas of specialized intestinal metaplasia (SIM) or dysplasia in Barrett's esophagus can be identified at endoscopy, the number of biopsies could be reduced and the sensitivity of biopsy surveillance would increase. It has been suggested that methylene blue (MB) dye staining may be useful for this purpose. Methods: Nine patients were prospectively studied with Barrett's esophagus. Staining involved sequential spraying of 10% N-acetylcysteine, 0.5% MB and water. Quadrantic biopsies were obtained from Barrett's epithelium and collected in separate containers depending on whether they were taken from stained or unstained areas. Seven patients undergoing yearly surveillance were asked to compare the discomfort of this endoscopy with that of previous surveillance endoscopies. Biopsies were analyzed for the presence and the percentage of SIM and dysplasia by a nonblinded pathologist. Results: MB staining prolonged endoscopy by a mean of 8 minutes (47% increase in procedure time) and was associated with significant vomiting during the procedure in 2 patients. Staining was observed in all 9 patients. All 7 patients undergoing yearly endoscopic surveillance indicated more discomfort with endoscopy plus MB staining. Of 37 biopsies from stained mucosa, 20 contained SIM; of 23 from unstained mucosa, 15 contained SIM (57% sensitivity, 32% specificity for MB staining). Conclusions: In this small, nonblinded study MB staining was associated with prolongation of endoscopy, increased patient discomfort, and potentially serious complications and was neither very sensitive nor specific for SIM. It is our recommendation that this technique should not be routinely used in endoscopic surveillance of patients with Barrett's esophagus. Further studies of MB staining are needed. (Gastrointest Endosc 2001;53:333-5.)