Intestinal metaplasia in the columnar-lined esophagus (CLE) has long been recognized as the most significant histologic risk indicator for esophageal adenocarcinoma. Recent concern has been expressed, however, that nonintestinalized metaplastic columnar epithelia (cardiac epithelium in the esophagus) may also indicate risk. Of 2586 consecutive patients undergoing endoscopy and biopsy in the Foregut Surgery Department, we selected (a) 214 patients with a visible CLE who had systemic 4-quadrant biopsies at 1 to 2 cm intervals, with the most proximal biopsy straddling the squamocolumnar junction, and (b) 109 patients without systematic biopsy who had dysplasia or adenocarcinoma. In the first group, 187 (87.4%) patients had intestinal metaplasia, and 27 (12.6%) had cardiac epithelium. Dysplasia or adenocarcinoma was present in 55 patients, all with intestinal metaplasia; its presence was significantly higher than in the cardiac epithelium group, none of whom had dysplasia or adenocarcinoma (P=0.01). In the second group with limited sampling, 49 had only tumor tissue in the biopsy. Of 60 patients with nontumor epithelium, only 34 (56.7%) had residual intestinal metaplasia. We conclude that systematic biopsies of CLE as described in this study separate patients into those with and without intestinal metaplasia in such a manner as to remove the possibility of false-negative diagnosis of intestinal metaplasia. When intestinal metaplasia is absent using this biopsy protocol, the patient is at no or extremely low risk for dysplasia and cancer. When biopsies have a lower level of sampling of the segment of CLE, the absence of intestinal metaplasia cannot be interpreted as a true negative for intestinal metaplasia. Inadequate sampling is a powerful reason why the near absolute association between intestinal metaplasia and adenocarcinoma is not seen in some studies.