Currently, the American College of Gastroenterology requires identification of goblet cells in mucosal biopsies from the esophagus to diagnose Barrett esophagus (BE). Identification of goblet cells in mucosal biopsies is fraught with limitations such as sampling and interpretation error. One previous study by our group suggested that MUC2 expression in esophageal nongoblet columnar cells represents a late biochemical reaction in the conversion of mucinous columnar cells to goblet cells in BE. We conducted this study to evaluate the prevalence, sensitivity, and specificity of MUC2 positivity in nongoblet columnar epithelium for detection of goblet cells in the distal esophagus and gastroesophageal junction (GEJ) region. We also sought to identify associations between MUC2 positivity and clinical and endoscopic risk factors for BE. This analysis utilized mucosal biopsies of the distal esophagus or GEJ from 100 patients who participated in a community clinic-based study of patients with chronic gastroesophageal reflux disease evaluated prospectively in the western part of Washington state. We randomly selected 50 patients who had columnar epithelium with goblet cells, representing the study group and 50 patients without goblet cells, representing the comparison group. Immunohistochemistry for MUC2 was performed on samples in a blinded manner without knowledge of the clinical or endoscopic features of the patients. The presence of staining was noted in both goblet and nongoblet epithelium, both close to and distant from the mucosa with goblet cells, when the latter were present. All study patients showed MUC2 positivity in goblet cells. MUC2 was present in nongoblet columnar epithelium in 78% of study patients with goblet cells, but in only 4% of controls without goblet cells (P<0.0001) (sensitivity, 78%; specificity, 96% for goblet cell metaplasia). MUC2 was significantly more common in nongoblet columnar cells close to, rather than distant from, the mucosa with goblet cells (P<0.00001). Finally, MUC2 was significantly associated with endoscopic evidence of columnar metaplasia in the distal esophagus, and with known risk factors for BE, such as older age, white race, frequent heartburn, and elevated body mass index. We conclude that goblet cells likely develop from a field of MUC2-positive mucinous columnar cells, and as such, MUC2 represents a late event in the development of goblet cells. MUC2 staining in nongoblet columnar cells is a reasonably sensitive and highly specific marker for goblet cells in the distal esophagus and GEJ, and its presence is predictive of endoscopic columnar metaplasia of the esophagus, even in patients without goblet cells.