Background: The diagnostic advantage of methylene blue (MB) chromoendoscopy in Barrett's esophagus is unclear. Methods: Patients with columnar-lined esophagus (CLE) were enrolled into a prospective, randomized crossover trial of MB-directed biopsy versus conventional biopsy. Results: Forty-seven patients (19 long-segment CLE; 28 short-segment CLE) were enrolled and underwent MB-directed biopsy. Sensitivity and specificity of MB for specialized intestinal metaplasia were 53% and 51%, respectively. Sensitivity and specificity of MB for dysplasia were 51% and 48%, respectively. Thirty-five patients (15 long-segment CLE; 20 short-segment CLE) completed the crossover trial. Relative frequencies for specialized intestinal metaplasia were 73% and 71% from MB-directed and conventional biopsy specimens, respectively (p = 0.73). Relative frequencies for dysplasia were 20% and 18% from MB-directed and conventional biopsy specimens, respectively (p = 0.65). In patients with long-segment CLE, dysplasia was diagnosed in 10 patients with MB and 7 patients with conventional biopsy methods (p = 0.25). The number of biopsy specimens per EGD was greater with MB, which may have influenced the diagnosis. Histologically, the grade of dysplasia was indefinite/low in nearly all of the dysplastic specimens. Conclusions: Results of MB-directed biopsy were similar to conventional biopsy in detecting specialized intestinal metaplasia and indefinite/low-grade dysplasia. MB was not useful in short-segment Barrett's esophagus. (Gastrointest Endosc 2001;54:294-301.)