Immunofluorescence staining for carcinoembryonic antigen, secretory component, and epithelial IgA was evaluated semiquantitatively in routine formalin-fixed mucosal biopsy samples from five patients with ulcerative colitis who had undergone colectomy because of carcinomas. Selected areas were given fluorescence intensity scores without knowledge of whether dysplasia or reactive hyperplasia was present as judged by another observer from conventional histopathological features in adjacent sections. The two types of lesion did not differ significantly with regard to the expression of the three marker antigens. In a prospective study based on cold ethanol-fixed mucosal biopsy samples, lesions from 11 ulcerative colitis patients with dysplasia were compared blindly with lesions from six patients with reactive hyperplasia and with samples obtained endoscopically from eight normal controls. The range of disease-associated fluorescence intensity scores was wide, but staining for all markers tended to be brighter in reactive hyperplasia than in dysplasia (P less than 0.01). In the controls, the fluorescence intensity score tended to be lower for carcinoembryonic antigen but was significantly (P less than 0.01) higher for secretory component and epithelial IgA than in both types of lesion. Moreover, staining for secretory component and epithelial IgA in the lesions seemed to be inversely related to the grade of dysplasia and the degree of inflammation. No such trend was seen for carcinoembryonic antigen. The wide ranges of individual fluorescence scores precluded the possibility of applying carcinoembryonic antigen, secretory component, and epithelial IgA as immunohistochemical markers to differentiate between dysplasia and reactive hyperplasia in routine diagnostic work.