The methodological issues for measuring colorectal epithelial cell proliferation, an intermediate end point for studies of colon neoplasia, in epidemiological studies are deceptively numerous and complex, with few methodological data available. Accordingly, during our experience with measuring colorectal epithelial cell proliferation from nearly 500 participants attending over 1300 study visits over a 6-year period, we recorded data on a variety of measurement variations. Methods investigated included rectal biopsy technique, general histological and labeling procedures [including the tritiated thymidine, 5-bromodeoxyuridine (BrdUrd), and the proliferating cell nuclear antigen (PCNA) immunohistochemical techniques used to label S-phase cells in colonic crypts in rectal biopsy specimens], biopsy scoring procedures, and summary scoring methods. Findings include that the PCNA technique was the simplest, most economical, and least time-consuming. The BrdUrd labeling failure rate was 15% versus < 1% for PCNA. The percentage of labeled cells (labeling index) was highest using PCNA in biopsies processed without prior incubation, intermediate using PCNA in biopsies processed with prior incubation as for BrdUrd, and lowest using BrdUrd. The percentage of labeled cells that were in the upper 40% of the crypt (phi h) was higher using BrdUrd than PCNA; visit-to-visit correlations were higher using PCNA (r = 0.51 versus 0.35), and visit-to-visit variability was lower and between-person variability was higher using PCNA. Intra- and inter-rater reliabilities for the techniques were comparable (PCNA intra-rater r = 0.93, inter-rater r = 0.92). The PCNA technique, compared to the BrdUrd technique, is more feasible and reliable, provides a more accurate estimate of the labeling index, and cell proliferation measures determined with PCNA have statistical properties that are generally more favorable for detecting differences in clinical trials. Thus, the PCNA technique may be preferable to techniques requiring incubation of biopsies. Other methodological findings lead us to recommend that, for larger studies measuring colorectal epithelial cell proliferation on outpatient rectal biopsies, biopsies should be taken 10 cm above the anus using a flexible, preferably jumbo cup, endoscopic forceps through a rigid sigmoidoscope, and histological sections should be 3 microns thick taken 50 microns apart.