Rectal mucosal proliferation has been promoted as an intermediate marker for risk of colorectal neoplasia. Proliferating cell nuclear antigen (PCNA) immunohistochemistry has become a standard method to measure cell proliferation. Whole-crypt dissection may provide a technically simpler method for determining proliferation within an entire crypt. We conducted a study to assess the reliability (reproducibility) of whole-crypt dissection in 10 subjects. Reliability of whole-crypt dissection with the subject as the unit of observation was excellent. The intraclass correlation coefficient for subjects was 0.93. Biopsy-to-biopsy reliability was lower (r=0.86) and crypt-to-crypt reliability lower still (r = 0.35). There was poor correlation between measures of proliferation index using the two techniques (Kendall's tau = 0.13; P = 0.08). Compartment analysis based on the percentage of the total number of labeled cells appearing in each crypt quartile also did not demonstrate a significant correlation between the two measures. We conclude that PCNA labeling index and whole-crypt mitotic count are not comparable measures of rectal mucosal proliferation.